The AM Forum

In this thread

http://amfone.net/Amforum/index.php?topic=44480.0

Tom, K1JJ made me an offer I could not refuse, in conjunction with Phil, AC0OB who offered the perfect iron for the build at flea market prices. I had decided to build a 8000 modulated by a pair of 811's mainly due to the cost of the 813 tubes and the lack of a suitable 2Kv power supply. Both of these issues were now resolved, so why not build the big rig!

As we start out I am going to outline my thoughts on how I will approach this build, subject to the vast knowledge of this board to help me "get it right". I have most of the parts in hand or on a truck somewhere.

The phases will be as follows:

Physical design combined with power supply build
RF Deck build
Modulator build

I plan to eventually have an existing Digital VFO (https://www.theladderline.com/node/37) driving a 20W SS amplifier/LPF (QRP-Kits) to drive the RF deck. I may use my Yaesu 757GXII in the interim if I decide to start the RF testing prior to assembling the SS components. For the audio, I will build the mosfet driver that Tom suggested. I have not decided if I will lay it out on a PCB or build it as he did. I plan to drive the audio chain with a combination of RPi 3B+ using some open source audio DAW tools. My work here is documented in another thread.

So that is the preamble...

The physical layout will hopefully comprise a 6ft tall old buzzardly rack that began its life as a Motorola communications rack. It has a door and some nice silver louvers that will nicely dress it up. I just have to convince the plant guys at work that it will be better to let me have it rather than sending it to the crusher. This rack will allow me to put the transmitter components facing out and have the door on the back to avoid wayward cats/grandsons from being fried. I am not yet sure how I will place the supply components in the rack until I get some time to measure them all and see how to get them placed. I have not decided if I will paint it or leave it with its natural patina. It is that not so likable Motorola tan color.

My objective is to build this power supply in a way that I can disconnect the supply from this rig and directly plug it into my Drake L4B Linear that was purchased sans supply. I have the same power cable that Drake used that will allow me to build it to be electrically compatible (except hardwired for 220V only). That way I can just reach over and unplug the PS and plug it into the Drake using the Drake switches to turn it on. I know I could make it fancy and switchable but I really don't see the need and this is just so simple to do it this way. I have all of the Cinch Jones connectors. I am going to make one change - no Millen HV connectors. I am going to use RP-BNC high voltage connectors.

The power supply iron comes from a Gates BC-1T. I will have the transformer, the 8uf oil cap and the choke. I bought 2 more 32uf 4500V oil caps so will have a total of 72uf of filter capacitors. Gonna need a substantial soft start for that much capacitance. I have not decided on a design yet. I have the supply shown in the attached schematics, but will be re-drawing the supply so show the partitioning and component groupings. Going to also as a variac into one leg of the 240V to allow me to control the overall voltage.

Trying to decide if I should make a small control chassis to house the HV meter, switching and screen/bias supplies or if I extend the size of the RF Deck facepate and put the primary supply components there. Taking a look at the size of the screen power trans and the filament transformer, I don't think the RF deck has enough space and they have to be on top. The bias supply is small enough to be under the chassis. Even using my planned 17x17x4 chassis, there is just not enough space I fear. Have to think about that.

Tom, I need to see a pic or get a description of the diode block. I plan to use 15 of the 6A10 diodes per leg. I would like to see how you laid these out and the use of the fuse wires that have protected the strings from disaster as you wrote in some of your threads. I assume it is on plexiglass on ceramic standoffs but the specifics are not firm in my mind.

John

No sense in cramming things in making it hard to work on.
A big transmitter should be big, just as happy chair is happy.

John,

Sounds like a nice plan!  Take a lot of pictures and post them.

It will be fun to watch the rig come together. I am excited that you will be trying to duplicate the RF  layout of Chuck's rig.  Your rig will have evolved by 2-3 builds. IE, it would have taken you several homebrew rigs before you discovered and employed these quality parts and advanced homebrew AM techniques on your own.  You have covered everything from the MOSFET audio driver, computer audio processing, digital RF drive, stiff HV supply, excellent RF parts and layout, commercial iron and big modulators - big audio to RF ratio... nice frickin rig!

I especially like that you listen to advice from others here and actually go out and do it. A more stubborn ham (like me) will do some things his way and regret it later...  ;D


The HV diode stack:

It's just a rectangular piece of Plexiglas mounted on four ceramic pillars.   Some guys use green G-10? fiberglass which looks very professional.  The diodes are mounted in two 15 diode lines. Drill 30 holes  for 15 diodes  and another 30 holes for the other 15 diodes.  Connect them together and solder the leads.   You can lay them out any way that looks pretty as long as the two legs are spaced apart by a couple of inches to avoid dust arcing, etc.    I like the look when the individual diodes are laid side by side instead of in a line.  

The HV fuses:  Big commercial vacuum HV  fuses are expensive... so I mounted  two ceramic pillars spaced 2"-3" apart, and at least 2" high.    Strip a piece of RG-213 shield wire (the outer braid) or equivalent. This is very fine wire. Place 1 to 2 strands in parallel to bridge the gap of the two ceramic pillars.  Use one fuse in series with the stack junction -  at the fullwave diode output just before the filter CHOKE.   You are using choke input so this will give some surge reduction. This assumes you decided to put the choke in the + lead, not the xfmr CT as mentioned earlier in a post.

Bear in mind when a fuse blows, the wire will probably flame an arc a few inches around it and below it... and above it. Mount the fuse in the clear and away from the diodes!  

You will know how many strands to use by experience. Start with ONE strand and if it blows from normal operation, then add another strand.   Bear in mind that these fuse wires can fatigue after a lot of HARD turn-on surges, so if one blows during normal operation after a few months, this probably means to add another strand.   You COULD use the wire/current tables to estimate wire size needed, but trial and error works just as well.  Your step-start will be important for your planned 72 uF of capacitance. A very slow step-start may mean no fuse fatigue problems at all.

The HV fuse definately works. I've had direct shorts for various reasons in the several rigs it feeds and never lost a diode stack in at least 25 years now.  But when the HV fuse does blow, it will leave a bit of soot on the pillars, depending on the intensity. Generally, the 240 breaker will pop at the same time as the HV fuses blows. Sometimes the fuse blows and the 240V breaker  does not.  This is what you want to see.  Coupled with the 25 ohm HV series resistor, we have an overall failsafe system.

BTW, my 4X1 rig HV supply uses three strands of RG-213 braid and has remained intact for a few years now. (140 uF @10 KV, 30 A choke)   Probably cuz I haven't been on the air...  ;)


T

Thanks for the diode stack info. Got it laid out on paper. Will get a piece of plexi this week and get it built.

I have decided build a power control center. Here is how it will work.

240v is supplied to the HV unit and up thru the control cable to the cinch-jones connector. It goes into the L4B amp or control head then back out to the supply as switched 240v. I plan to put a relay in the HV unit to bypass this current path so that the voltage going back is low current. This is a common update for L4B supplies to allow for reduced current on the switches. This will also insure that I am not carrying full current for the trans up and down the cable. The L4B transformer is 700VA. This Gates one is at least 1500VA. The 240 will have circuit breakers and soft start in the HV unit.

Should I have the control head get a separate 120V or should I send a neutral up the cable and split the filament and supplies across the load and run it all off of 1 240v feed? The schematic shows sending ground on pin 7 but no neutral. Pin 8 is not connected in the L4B supply but it is grounded on the L4B amp chassis. I could lift that ground and pass neutral on pin 8. My house is build according to the 4 wire code so my neutral and grounds are separate except at the transformer pole where they are bonded together.  

I think this will be safe. My reasoning is that I am the one that will be using my HV with my amp. There would be concern if someone took my HV supply and plugged it into another amp. The concern being that the neutral and grounds would be bonded at the amp. My amp being plugged into another L4B supply would be no concern at all, I believe.

Thoughts?

The control head will have fused switching for filament 120V. I plan to use those nice computer power connectors to feed the 120 voltage out of the control head up to each chassis (mod and rf) to the filament transformers. I will have 2 RP-BNC connectors to loop the HV from the HV unit and feed it out to the mod deck. Then 2 more connectors to loop it from the mod deck to the RF deck. I can buy pre-made HV cables with nice insulated ends. I will also have an HV meter on the control head so there will be a resistor string there to drop the voltage to a measurable value. Variac on one leg of the HV input feed to control overall HV output.

The fused screen and bias supplies will be built entirely on the control unit chassis, including the VR-75 tube and small variac. See schematic for locations where the split occurs (the split at VR75 is not correct on the schematic). The meters for these supplies will be located on the RF front panel.

I plan to have a BIG RED SWITCH on the primary 240V feed mounted prominently somewhere that can shut the whole mess down. I don't want to have to be reaching to unplug something if the main relay decides to fuse closed or some other major failure.

Well, I think that is the plan so far. I will revise my schematics to reflect this partitioning plan. I like my schematics to be a good roadmap for the build and a good reference if I have to go back 5-10 years later to repair something.

John


 

I always use the HV fuse of magnetron ovens, they are ok upto approx 3 kV, small and enclosed in an isolating cartridge. And when they blow, i do the same as you did, use thin wire to repair the fuse (if you don't have spares or a shop closeby)

Is this going to be a push pull circuit? I built an amp using 8000's way back around 1978 and it worked great in push pull. It had more output than 810's in the same circuit. 8000's are quite rare so if you run out of 8000's, 810's are not hard to find and will most likely other than bias adjustment, will be a drop in replacement.

Is your power supply going to be cap input? That's a lot of capacitance and could result in an electrical explosion if something were to short out.  The link below has fuses and holders.

https://www.ebay.com/sch/i.html?_from=R40&_trksid=m570.l1313&_nkw=Microwave+Oven+High+Voltage+Fuses&_sacat=0 (https://www.ebay.com/sch/i.html?_from=R40&_trksid=m570.l1313&_nkw=Microwave+Oven+High+Voltage+Fuses&_sacat=0)

There are other fuses listed at higher ratings in amperes by looking around eBay but most are rated at 5 KV.

I cannot express forcefully enough to make every effort to make sure that your capacitors are discharged completely and that a ground shunt or short be in place at many points where lethal voltage may be present! Just shorting something out with a screw driver is not enough! The shorting mechanism need to be in place throughout while high voltage repairs are being made!

Points noted on the danger. Power supply is choke input as shown in the attached 813 schematic.

Do you have an example of a shorting mechanism that is safe when in operation?

Here is what I am building - http://www.amwindow.org/tech/htm/813/813.htm

I _was_ building an 8000 rig...

The Collins KWS1 uses a leave spring that shorts the HV as soon as the top plate is removed. Very simple and safe system. If someone has a KWS1, he can make a pic of it.

Commercial systems use an interlock system. I'll get back to you with an idea or two sometime tomorrow. Too tired tonight to strain the brain  ;D ;D

I guess I need to read more closely to understand what you are doing? All I had to do is read :D :D

I use perf board from....  RadioShack.com for my diode stacks.

Less drilling 😇

--Shane
KD6VXI

I suggest a 4 wire 240V cable to the power supply, then run 120V, neutral, and GND to the controller from a jones connector etc. on the power supply. You can be sure it's all un-powered whenever you disconnect the single 240V plug, and since the control has a special plug, it can't be plugged into a 120V outlet by error.

The reason I suggest this is because I used to run a separate 120V power cord to the control box and got a very nasty surprise while working on the unplugged Tucker transmitter, as the station control was still plugged in and supplying 120V to the transmitter in a roundabout way. The worst part was not the shock, but the large cut on my arm as I jerked it out of the bowels of the transmitter.

First thing you know John is that next you'll be feeding the output of the 813 into a setof parallel 833s or 4-400s. ;D

Phil - AC0OB

Is there no end to this madnesss?  ;D

Great suggestion, Patrick. I will incorporate it. That way no mods to the L4B.

385 pounds of iron arrives today  8)

And six pounds of 813s, caps and sockets are on the way to you today now that the ice storm is over..

The postal worker wanted to know if I wanted to pay an extra $10 for a "fragile" sticker.  I said no and  figgered it was packed well enough with styro and bubble pack so that the tubes would be FB.    813s are possibly the most rugged big transmitting tube ever made.

So what's the latest on the project over the last week?

The most difficult part may be finding a name for your rig. I've had rig naming contests here in the past.  Just so you know..."Fabio", "Dr. Love",   "Dual Quads" and "Rico Suave"  are already taken. Or you could do like most hams and just call it "The 813 AM Rig."     ;D


T

Thank you Tom for keeping your promise on this deal and your help along the way. I would not have taken this leap without support.

Going to do the physical layout of the power supply today, and build the diode array. I assume I should be using the center conductor from some poly RG58 for all voltage interconnect on the supply?

Have electrical connectors showing up to do the interconnects to and from the control panel. Also going to lay out the screen/bias transformers and power supply circuits on the chassis. I have a free-standing supply for the 24v relay power.

Want to thank Phil for delivering the iron. As we were talking in the parking lot, for a fail-safe Phil suggested a large relay that is energized from the line power that holds off a large resistor when on. When line power is removed it is placed across the caps to quickly drain them and keep them shorted. He suggested a 200W one with sufficient resistance to bleed it off quickly. What can I do to reduce the arcing on the relay - some decoupling caps across the relay will not be enough. Going to do a little googling to see what others have done. Guess I could us one of those nice russian spst vacuum caps here as well?

Thoughts?

Also will spend time this morning doing the final schematic for the PS.

Names? Gort is taken also. Was thinking "Lord please don't let this PS kill me" but that is too long.  ???

John

Here are the latest schematics. Reworked the power so it shows the HV unit vs. the power controller. Still have to design a soft start. Suggestions for a KISS approach?

20A or 30A breakers? 30A or 40A main breaker on the panel? Guess I should check the Gates schematic to see what they did.

On edit - the Gates fused at 30A providing 3500W of power at 85% modulation running a pair of 833s driven by a pair of 833s. That works out at ~15A steady state. This transmitter should pull about the same or somewhat less, I would guess. I think I will have breakers at 25A on the HV unit and a 30A electrical circuit. Soft start will help here on the surge current.

Phil gave me the Gates filament transformer also. Looking to see if I should use it. It has multiple 10V filament feeds. What is the rule of thumb on how far the filament transformer should be away from the tubes themselves?

Added the antenna switching and the proposed sequencer order. Please help me here on the initial sequencing plan.

Is it necessary to have the 33K/200W additional R on the supply? It will drain the power at 80ma after the mains are shut off. That is 200W of power at 2.5kV.

Wondering if I should drop the 8uf oil cap. That was the original Gates cap. I have added 64uf of additional capacitance. Why bother with the 8uf can...

After this design session, I still have parts to buy. Thought I was about done... oh well..

John

Forgot the HV fuse...

I think this may be a non issue since you are bleeding off to a resistor and not an inductor.

What value of resistor are you planing on using? Bleeder resistor typically rage from 50 to 100 K depending on how stiff you want the regulation. I always suggest using resistors in parallel to give you a safety margin in the event that one resistor goes open. For example two 100k in parallel to arrive at 50k. 

Working on the step start part of the circuit.

My transformer is a big Gates transformer. Inrush current is going to be big given that I have 72uf of caps on the secondary. For safety reasons I decided that I will have good old fashion switches to cut off the power. They then feed to a relay where I can switch the HV from either the control head or the L4B Linyaaar amp. This relay is controlled by a 120v feed. After the relay I will have a variac on one or both legs of the feed depending on what I can find. Next in the line is the step start. I have read about them from various uses, mostly on amps with considerably smaller transformers. I think this one needs to be substantial.

I am going to build the step start along these lines. Maybe Phil can help me with estimating the specs on the transformer.

http://amfone.net/Amforum/index.php?topic=41787.msg303932#msg303932

I am going to use the 40H estimate until I can measure the trans. Not sure my little LC tester can go that high.
Caps - 72uf
Sec current - 1.5A
100K Bleeder

TC - R*C =7.2s (a long time)

R = L/TC = 5.55 ohms

Fuse = 1.5*2.5=3.75A

So looks like a 6ohm 200W R in series with a 4A SB fuse?

That TC is really long. If I go to switching HV during a QSO I think I will need a better solution.

Going to drop the 33K extra bleeder. I think I will use the existing 100K bleeder with a mechanical interlock to short out the filter caps if the case is opened.

New schematic. Going to start the PS build this afternoon if I don't get a hernia carrying the iron to the shack in the basement.

John

I've got that power supply in my shop.  Knocked me with 6kv.


--Shane
KD6VXI

I read the post - scary. That is why I am going with RP-BNC as suggested in a reply to that post. No Millen HV for me.

Glad you are ok.

John

Hi John,

Without going into formulas, I'll give you my practical experience on step-starts...

There are many ways to approach this. I personally like keying the HV for transmit/ receive. The goal is to reduce the initial major surge, that capacitor-only supplies can reach  … 10, 20, 30+++ amps, depending on the components and voltage level.   The diodes can handle this as well as most capacitors, but it does raise havoc on longevity if pushed to an extreme.

In your favor you are using a big choke input, which will oppose the flow of current as the mag field initially builds. This is a good thing.

Most of the initial charging occurs very quickly, in under a second.  A step start is the best way to add more "cushion" to the surge.  My suggestion is a 5-10 ohm 200 watt resistor that stays in the 240 VAC primary for 1/4 to 1/2 second and then gets bypassed - is about right for your supply.  That 1/4 to 1/2 second can get "buried" with the other functions of sequencer antenna/ driver switching, amplifier key up, etc.  The resistor contactor (and any other associated relays in the 240VAC primary) should be a robust 240VAC type, 24VDC coils, about the size of a lemon with heavy contacts, open frame... you know the type.  No RS 12 V relays.

You will see your AM carrier come up before the step start drops out and then there will a few DB power peak afterwards. You may hear the second click of the relay unless you use some kind of audio delay as I do, to remove all the shack noises on key up. IE, the audio comes on last.

If you make your step start timer variable, you will be able to find a compromise between key-up speed and reasonable current surge. I run this same setup and component values on my own 140 uF, choke input 4KV HV supply.  My total delay is about 3/4 second and the HV supply is barely stressed on key up.  The HV  B+ line fuses also provide protection and a good guide to how hard the system is being stressed.  If the HV fuses fatigue after only a few short days of use, then you need to either increase the fuse rating or decrease the surge startup level.
 
BTW, I will try to get to your other questions later... For now, your question about switching in the additional bleeder after shut down... you already have fixed bleeders on the filter caps, right? Then the extras are a redundant system for safety and is always a good thing if you want to go thru the extra building effort. Most hams don't bother, but if you are extra nervous, then do it.


No, the small 8 uF will just mostly take up space... stick with the two 32 uF beasts. Use it as a spare.


T

ok, SS is now variable using Square D octal timer delay relay driving PB open frame relays for HV on/off and step start resistor engage/disengage. Inexpensive and look robust. I have to run 120V coils however, since I do not have 24V in the HV unit. Schematic getting a bit busy in that part of the sheet.

I think this unit is now going to be separate from the rest so I can move it around to future transmitters and move it if I want to run the L4B. I am re-thinking the full 6ft rack. I may go to a half rack that holds the controller, RF Deck and Modulator. Still trying to decide. I don't have a lot of space in the basement so ease of moving is important. Stuffing the PS components in the bottom of the Motorola rack (if I get it) will make it very cramped and not very serviceable. That transformer and choke are HUGE. Thinking of a roll-about wooden platform with a removable interlocking metal cover that makes it easy to expose and work on the PS unit (power off of course). Kind of like a mover's dolly with a metal cage around it. I can then roll it under my tables to get it out of the way.

Trying to think of a indicator that will illuminate as long as there is potential on the caps. Or I will just mount a second meter on the HV unit. I want a visual indication of it being dead. Otherwise when I open the unit I will kill an interlock switch in the process.

John

Hi John,

Sounds good on the step start.

For relays: I would recommend staying with 24 VDC coils only - for all station switching.   They are generally quieter and safer than 120VAC coils.

I also have 12VDC available for T/R keying since it is sometimes needed too.

Try some of your AC coil relays and see if they buzz.  I ended up with about (15) 24 VDC coil relays  in my control system, including AC mains contactors. All those relays would be a lot of racket if AC, especially the bigger ones.  Some get mechanical resonances and buzz at 60 Hz. At least that has been my experience.  The first time I knew I didn't like 120 VAC relays was as a Novice switching the Johnson Matchbox ant relay. I blew a hole in the aluminum rear panel.... :-)

BTW, there are lots of cheap 24VDC small switching power supplies on eBAy, etc. I use a few here for 24VDC and 12 VDC. Ya never know when you will need another relay as you add equipment.

Basically I have a 24 VDC T/R keyed wire, running to all rigs. When a rig's panel T/R switch is on, that rig gets keyed and switches its antenna relay, mod and RF PA on, etc.   Simple.

T

Hi John,

Here is an inrush circuit that should handle the inrush current for initial charging.

It is a modification of the one that I use on my Henry 2k-4.


Phil

OK, all 24V now.

On edit - think I will slap a 2200uF cap across the surge relay coil and not put in the delay relay for now. That should give me around 3/4 of a sec of delay which should be a good starting point. I will make space for the third relay if I need to mess with the TC.

John

Relays are here, going to start construction.

Tom’s tubes are here as well!

So does the Heath SB220/221. Unfortunately, some have, without thinking, removed the cover with the HV on which usually ends up with some destruction of some HV components. Seems to me there should also be an interlock that would remove the primary to the HV xvmr along with the shorting leaf. No. No fuse will act fast enough to protect the HV components.

Al

PS: Forgot to mention that a "glitch fuse" works fast enough. W2DTC has some nice detail on this. Usually made of #30 wire but the Clipperton L uses a pair of carbon comp resistors that works well plus provides an interesting light show should they be violated.

Just check Ken's site and he prefers the light show so he knows that there is a problem. He also uses magnet wire
http://w2dtc.com/w2dtc-class-a-rf-amp/2007-1204-b-plus-glitch-resistors.jpg (http://w2dtc.com/w2dtc-class-a-rf-amp/2007-1204-b-plus-glitch-resistors.jpg)

Yes, there is also a microswitch swithing the HV off. I tested once and only the HV fuse was the victim, no damage done

If you place a cap across the relay coil, it not only delays the contacts making contact, but it will also hold the relay in for a while and delay the contacts breaking when you go back to receive.  To avoid this, you might want to add a diode in series with the capacitor, charging it while going from receive to transmit to delay the contact closure, but also allowing the contacts to close immediately without the capacitor holding the relay closed.  I used an extra set of contacts in my sequencer to discharge the capacitor through a resistor when the relay opens, such that the delay function is preserved the next time the relay is energized. 

Yep, I was thinking the same thing, Rick, but forgot to comment on it.   Before I went with all adjustable delay relays for the sequencer controls, I tried the capacitor/resistor/diode circuit. It worked OK, but it took some messing with to get it perfect.   I've found as you change things and want different timing responses in the shack, the timer relays are a cinch to make things right with a turn of the knob.  

Somewhat elaborate I admit, but with mine, I can independently adjust the attack and release times and keying order - which helped to fine tune my arcing T/R problems with the 4X1 rig. It also controls HV step start, antennas, mod deck, RF keying, audio, mod reactor quenching, etc.

There is a sweet spot in adjustments where the transmitter keys and operates smoothly and near perfect..

Page 8, 9 shows and discusses the sequencer details. It was designed by Jeff, W2NBC:

http://amfone.net/Amforum/index.php?topic=33816.200


T

FB on the adjustable delay relays, Tom.  That certainly is a convenient way to fine-tune the sweet spot.  

I designed an RC time constant sequencer, amplified by a few cheap NPN transistors to drive the relays, thus taking the relay coil variations out of the equation.  I put these on a perf-board under the chassis in my Viking II and Valiant.

Currently I am finishing up the software to do all sequencing with an Arduino Nano ($3) and a relay module with eight optically-isolated SPDT relays ($6) and a 4-line 20-character LCD display ($5) with a couple buttons to configure the desired options.  This allows me to keep several transmitters attached and select the desired combination via a simple menu.  The buttons were two for a buck and the entire shebang is powered by a 5-volt micro-USB cellphone charger.  Can't get much cheaper than that!

Don't want to hijack John's 813 thread, but knowing he is a software jock, this idea might interest him as well.  Real simple to code in C with the Arduino IDE.  Cheaper and simpler than a bunch of RC circuits!

Got my interest, Rick.

There are 2 issues at play here and I am only trying to solve one now. I understand the need for overall sequencing (Fabio II was very illustrating) and the strong desire to have it quite flexible.

The primary engineering issue at hand here is how to safely handle the inrush current on this very large supply properly so that every time I activate it the diodes are protected while those large caps charge. I think this needs to be a fixed TC and absolutely independent of overall sequencing or I will risk PS components and safety. I can't let a mistake or failure on sequencer delay cause the HV PS to blow fuses or arc.

This also under the constraint of having to use one lead of 240v on the input to energize the supply to maintain compatibility with the L4B. 24V coils makes that much more complex.

After looking at all the options, I think I will try Phil's approach. I plan to have the overall HV controlled by a 24v line on the controller that I can drive from the sequencer. And if this idea that Rick has looks good, I will have excellent overall control of the rig.

I also have a mechanical interlock switch that will hopefully only blow the fuse if it is accidentally tripped.

New schematic.

John

Since we are also talking about sequencing - I read the Fabio II thread. What would a good starting approach be? (Ignore my sequencing in the schematic) I have relays for HV, RF drive, PTT (mod and RF Deck keyed at the same time), and antenna. I have not made a provision for screen.

I am thinking about separating the RF Deck and Mod PTT relay mainly because I don't want to run a line from the relay on the mod deck across to the RF deck (see schematic).

Key on

1 - Antenna
2 - PTT (both Mod and RF)
3 - HV
4 - RF Drive
5 - Audio

Key off reverses.

thoughts? Should mod and RF deck be on separate sequences?

John

This is great.  I am following along on your project closely.  I have wanted to build the K1JJ rig for several years but am still on the hunt for the last remaining piece - a dang modulation transformer.

-Steve

Hi John,

Yes, Rick's adjustable delay software sequencer will suit you fine.

The sequencing timing you show above is a good starting point.   I use a screen relay in my sequencer keying on the RF deck.  Your mod tubes in triode config do not need screen keying IIRC.  When you test you will see.  My own 4-1000A modulators are tetrode config and do use a screen relay and regulated supply.

I believe I made my TX key and mod key independent.  Reread the Fabio sequencer thread on pages 6,7,8,9,10 to see what I went thru.
There were a lot of changes and experiments before arriving at the best optimization!
http://amfone.net/Amforum/index.php?topic=33816.200

Once running, there may be small screen/grid/plate  meter jumps on key up and unkey that will require some sequencer fine tuning.(assuming it's not caused by parasitics)  I even changed the key up order at times. You will have to test and see.

Bear in mind that some big rigs have no sequencing at all and get away with it FB.  A sequencer is simply another step to making sure the rig operates smoothly and without surges, arcing, etc.  I always use a sequencer for all big rigs here and gain full control of the rig.

Another thing to consider adding is a 5K-7K 100W power resistor that gets switched across the modulation reactor during unkey to quench the arc and dissipate the magnetic field. It gets keyed open during transmit using a vacuum relay. This makes a big difference when it's time to unload the high power audio and protect the mod iron..   Normally the antenna takes the RF deck dumped energy without this kind of relay circuit, but the mod xfmr and mod reactor are not always able to dump it easily into the RF deck and may create spark gap arcs that can be excessive.  During sequencer testing I seem to remember keeping the RF deck on for an instant while the modulator finished dumping its power.

Anyway, you may and may not need these extra features, but now you know what to do if these problems come up.  Think in terms of "where will the dropped power go and what currents will want to soar when I key and unkey?" "Will a different order and timing help solve these problems?" "What will happen if I turn the screen on without the HV?"  "Is it best if the audio keys on last so that the whole modulator chain and RF deck have good loads?"  "Will the RF deck screens stay in safe cutoff if the RF drive comes on before the HV?"    These are good questions to learn from and get you thinking in the right direction.

T

Hi Steve -

Welcome with your first post!

Iron is usually no problem. Put out a "wanted" ad on several of the web ham bulletin boards for broadcast modulation iron, (mod xfmr and mod choke)  500W - 1KW range.  (like QRZ.com, etc.)   There are many old ham AMers - some from the 50's and 60's in its heyday, with this heavy iron in the cellar who will never get around to using it. They'd love to have an enthusiastic homebrewer take it away. Got a strong back?  :-)

Write in the ad you are building a homebrew 813 ham transmitter for 160-20M. (or whatever)  This will inspire the OTs to help you out...

Try and see. I'll bet you find some quickly.   John had the same problem and found it in only a few days right here on this BB.


73,

Tom, K1JJ

Hi Tom.   I am actually not new here.  For some reason I was no longer able to log in with my old user name "stevef" so I didn't see any option but to create a new account.   Anyhow - I will take your advice - Thanks!

Steve   
 

Still ironing out a few details. I was not happy with the mains switch setup. Now have a 3pdt relay with a start-stop no/nc switch. This will give me an auto power off if there is a power glitch. Parts still not all here so no progress on the actual build other than rough layout. Main item to fit just arrived on Monday, a Superior Powerstat. Going to have ability to vary HV from around 1500v on up.

Watched the ARRL video of the Gates bringup with Tron. Noticed they were running the HV at 1800v. Was cool to recognize the same components that I have.

Question on filaments. I have 2 separate transformers or I can use the dual 10v one from the Gates. Is it better to have the filaments trans close to the tubes it feeds or can I mount the dual one and feed the tubes with 2-3 foot leads?

Filament transformer leads need be large enough to avoid voltage drop. Otherwise no issues with a remote transformer. I presume you will monitor filament voltages at the socket, so all should be well.

Which MOSFET audio driver are planning to use?

John,

Yes, as Pat says, any length filament leads will do, but if they are too long you may not have enough tap or Variac range for  the fil xfmr to cover the voltage drop. Just a waste of power since 813s take a fair amount of fil juice.

Just be sure to well bypass the RF stage 813 with .01s at the fil pins... and also bypass the modulator tubes at the fils with both audio and RF bypass caps. I see no audio bypasses on the Tesla 300 schematic.   I have run tests without audio bypass caps and did not see much difference in sweep performance, but would hate to think that the extreme highs were getting thru the iron without problems. If anything, bypass both stages well to enhance stability.

Yes, you can use either separate fil transformers or dual windings. Once the RF or audio gets bypassed to ground, the stage doesn't care where the 10 VAC comes from.

T

How are you now BTW? I hope healed up well!!

For low voltage high current filaments a good and very flexible wire is the stuff used by car stereo shops for high power amps. It's made of many fine wires like a welding cable (The #10 may have 500-1000 strands), instead of a few coarse ones like 'mains' wiring. It's cheap too.  Just encouraging overkill..
Its insulation is usually silicone type and not as tough as the 600V stuff, but heat shrink tubing helps.

The one that Tom used on Fabio II.

John

This thread:

http://amfone.net/Amforum/index.php?topic=23632.0

(Schematic on webpage above, 3rd post.)


T


Prototype and current PCB:

Wasn't there a picture of the power supply of this, like a short rack with a quite large size power transformer sitting in front of it?

I'm back working in the shop.  Cost me about 6 weeks of downtime but better than the alternative!  Dirt nap!

Glad you got the transformers.  Those oil caps, not sure I'd ever use them so figured I'd send them to someone who may.  And I still have a bunch left over!

If your asking me about the short rack power supply, no.  That was a different supply I built for someone.

The amp / supply that got me was built by someone else.  Incidentally, I've received another amp with a Killen connector.  Same problem.  Wire abraded going into the connector.  Idiots.

--Shane
KD6VXI

While I wait for the last relays to show up, I am going to knock out a pcb for the driver. Was never able to get hold of the guy that did the last one. Should not be too hard.

John

Separate thread for the Mosfet driver board. I am laying out a new PCB.

http://amfone.net/Amforum/index.php?topic=44619.0

John

OK, I just thought that I might have the same transformer as in front of the short rack power supply, and wondered if it was used with a bridge. IIRC mine has three big ceramic terminals on the HV side and lightly stamped '5800' but I have to check. (of course it is on the bottom shelf with the business end facing away, and it's too heavy to be moved easily..) Also wondering what the transformer is from. I guess I should post a picture of it and ask in a separate topic, kinda mixed up as to where on here I saw it, thought it was in this 813 build topic but can't find it now.

Well as far as those connectors - if you insist on correcting that whenever found, future generations of people may live to thank you.

I came across this old pic of my Plexiglas 4-1000A modulated by 833As.  I had this thing about  being able to view the insides cuz we soon forget what it looks like once in a steel cabinet.  It worked pretty well but got transformed into Fabio II, the latest 4X1 X 4X1s. (in pics #2 #3)

Notice everything is on wheels or a pedestal - and EZ slip off enclosures for acccessible servicing.

A good excuse to check up on the progress of John's latest 813 rig...

T

Ordering a chassis along with the parts for the impending mosfet board arrival and going to start metal work layout on the rf deck. Have been conversing with Chuck to get the placement of the right angle turns and turns counters correct. My tank coil and band switch is huge so I plan to use a 17x17x4 chassis so I have plenty of room. I also picked up a La Pointe roller for the reverse Pi grid input. Now I need to find another multi-turn apparatus. Thinking of a vernier knob for this one. Needs to be smaller to fit right.

The buying never stops, it seems.

Where do I insert the grid bias in the filter? on the cold side of the roller? Where do I use a cap to block the DC - assume on the input to the reverse side and I should use a 500pf value? Do you perhaps have an example schematic?

Should I be planning on forced air flow into the chassis bottom and plan to make openings around the tube sockets to get air flowing up around the tubes?

John

Hi John,

I'll describe the grid circuit and then you draw the schematic and we'll go from there...



There are three paths, all starting at the 813 grid pins:


1) 2K > .001 cap to gnd

2) 2.5mH > .001 cap to gnd.    From .001 cap hot to grid leak resistor ... other side of GL >  .001 cap to gnd.   Hot side of cap >   -75V reg bias.


3)  .01 cap series >  300 pf variable to gnd.  Hot of 300 pF > roller inductor in series.  Roller other end> 1200 pF variable  to gnd.   Hot of 1200 pF > .01 coupling cap.   Other end of coupling cap is RF input.


No need for neutralization because the G-P is only 0.25 pF per tube and the 2K> .001 to gnd lowers gain to stablize everything. Good layout will help further.   Many have run 813s without neut with success.


Air cooling: Several possibilities:  Use heat-proof glass latern chimneys with holes around the socket. Squirrel cage pressurized from back or top of chassis. -  Underneath air pressure cools fil seals and glass envelope on top.   If no chimneys, then holes alone will suffice, but top plate seal is not cooled well unless a muffin fan directs air down from top at center of tubes in addition to air flow at bottom to cool fil seals.


If you can locate the proper size chimneys on the web, it makes a HUGE difference in overall cooling. I did that here and the 813 tubes show their appreciation by losing their blush... :-)

T

I think I am 100% complete on the design, Tom. Check closely the latest grid changes.

I have incorporated all that I can think of and all of the great ideas offered on this board. The project scope is huge but the benefits will be amazing... Think about where this started with a simple 1700v, 350ma supply and a single 8000 modded by a pair of 811's. Quite a difference!

Items to note on this revision -

1. New Grid input circuit with no neutralization planned - will build it very tight and shielded from the output tank. Floating roller inductor (found a nice one!)
2. Mosfet board interconnect added with NFB tap
3. Separate control of Mod and RF CT's
4. Switchable RX only antenna - going to try a LOG on 160M - http://www.kk5jy.net/LoG/
5. Arduino sequencer board - 6 steps planned (need to design this)
6. Safe start PS input - will shut down automatically if the power blips
7. Frequent use of Russian SPDT Vac Relays (They look like Plankton from Sponge Bob Cartoon!)
8. Spark gaps (will need to finalize a design and placement)
9. Changed to use the Gates dual 10V filament transformer.

I have all of the parts except the chassis and relays. I think I am finally ready to start! (measure twice - cut once analogy appropriate here)

Paper is easier to change than metal.

My PS will be started later. I am still mulling over the final enclosure. The rack that I might get from work only allows a 17in by 17in space. That is not enough space to cram all of that Gates iron into it without a lot of KY Jelly. I don't want to cram it in and have arching issues if a darn spider decides to make a web inside. A Gates enclosure is quite spacious in comparison.

I am going to start on the RF deck first. It will be on a 17x17x4 hammond chassis with I think a 10U front panel, with right angle drives underneath to the turns counters. The vac caps will be mounted vertically on either side of the chassis. Two tubes up front, with chimneys. Large Tank Band Switch and inductor behind the 2 tubes with the switch rod running between the tubes. Input tank vernier and knobs below.

Face will have meters across the top, tube viewing oval holes with the TBS knob in between. With chimneys, may skip the holes since the tubes will not glow much. Input tank controls south of the BS in the center, Plate and load outside on the same axis as the input tank.

I think that is it! Any comments or suggestions will be readily accepted.

John

Looks good, John...

Just a couple questions regarding your schematic....

The plate current meter in the 813 Final cathode (filament) circuit will read combined plate, screen, and grid current.  You may want to consider floating the positive return of the control grid supply, as well as the negative side of the screen supply, and returning them to the center tap of the filament transformer.  Alternatively, you may float the negative side of the high-voltage supply, and place the meter between the supply and ground such that the 1000 mA meter reads only plate current.

You might also consider a higher voltage rating for the blocking cap coupling the plate circuit to the output tank.  5KV seems a bit marginal.  I use a pair of Sprague 20DKT5 caps in parallel, 500 pF each at 20 KV.  One may be sufficient.  Consider the capacity in parallel with the output of your modulator and its effect (in addition to the plate supply bypass capacitor)  shunting the higher audio frequencies when sizing this capacitor.

I do not see any protection for screens, should you lose plate voltage or excitation.  Perhaps some logic that requires drive and plate voltage before enabling the screen supply, or maybe a clamp circuit?  If screen supply is present without plate supply, the 813s will be very short lived..  GL on your project.

John,


Hey cool.... this is my post 7777.   Lucky mega-jackpot!  Where do I pick up the cash?
;D

OK on all. That rig is certainly turning into a professional project!   Now if you can maintain the workmanship at the same high level, you will be in the rare category of having a rig worth MORE than you spent on it...   Unfortunately, our time is usually way below minimum wage.  ;)

As Rick said, there's no screen protection. The modulators are OK since they are triode connected, but the RF finals will be at risk. Some guys chance it and are careful, but I prefer SOME kind of scheme to protect us from dumb mistakes.  In my 4X1 rig, I used robotics industry magnetic sensors that could be adjusted for current to trip a latching relay that opens the PTT circuit.  The shutdown delay speed is not as important with tubes as it is with solid state. Tubes can take more abuse for a second longer, usually, so mechanical relays are OK.   SS devices are milliseconds or less to damage.  

I used these magnetic current sensors in series with each DC circuit for the plate, screen and grid circuits for both of the RF finals and tetrode modulator tubes. So now if one of these tube elements goes past my preset limits, the PTT circuit opens up. (the fil CT opens so no current will flow thru the tube at all) It's kinda slick. I could come in drunk, turn on the rig and not blow things up... ;D   Do a search for current sensors and see what you find. It's easy enough to design using the sensor and a latching relay which opens up the 24V keying PTT circuit with a set of contacts.

The screen clamper circuit will work FB too.  Yes, 5KV for the plate coupler cap is a little low. Go for 10KV or more.

Your grid input circuit looks OK on the new schematic.

BTW, you don't have to use a tube, VR75, for fixed bias grid reg. A SS 70-80V  10W+  zener will work too.  This regulated voltage is to keep the grid bias filter cap from charging/ creeping up due to RF rectification. When testing, be sure this fixed bias stays solid with a full RF carrier. If it rises, I have a simple two component fix if needed.


When using the 813 chimneys, the drilled chassis holes around the socket base increase the chimney airflow.  Another way is to use spacers to make a gap so that there is 1/8"-1/4" recess of the sockets. This way no drilled holes are needed and the gap lets the air thru. The tube is designed for no pressurized air required so anything you do is a bonus to cooling effciency. As long as some air passes by the fil seals you are OK.

You can monitor plate current either by mentally subtracting the grids and screen current (as I usually do) or do as Rick suggests and float things.  The plate current meter in the negative lead is FB too. It's all up to you.

I'll look closer at the schematic later, but I think you are getting close to a workable build soon. The others here will hopefully see some more improvements or mods to make.

***  The mod reactor cap (C1) should be more than 3KV. If you can find another cap, put it in series for 1 uF @ 6KV.


T

John, here is an example schematic (from Jun 56 QST) that shows how easy it is to place the plate current meter in the negative supply return instead of between the final cathode (filament) and ground....  All of course better than the old way of putting it in the positive plate supply lead - (ouch).  The 5 to 10 ohm 50 watt resistor ensures the negative supply connection is not broken, even if the meter is disconnected or fails.  The resistance is high enough that it will not significantly affect the meter reading, even though it is in parallel with the meter.

It also shows properly-sized caps for the plate coupling and bypass.  Hope this is helpful to you.

I'll add, always but ALWAYS use a couple to a few 6A10 or 10A10 diodes from b- to ground.

They will clamp if that resistor opens up, preventing the chassis on the rf deck from approaching b+ value.

With a 400A surge rating they will take a lot more current than the meter and / or a 50 Watt resistor.

I size mine so the V Drop across them is the same as what is needed for plate current, then add one or two more to ensure they do not effect the plate current reading.

If you use inky enough diodes to pass said voltage, you start having the diodes conduct and pass current effecting the meter.

Try it on a small variable supply.  Two diodes in series will start passing current before 1.2 volts.

I've had to throw a LOT in series to repair / safe off others amps.  I had a 500Z amp come through here with all beautifully matches Simpson meters.  Except the HV plate current meter was in the b+ lead, causing 4kv to be present between the meter internals and ground!

Moved the meter to the b- side.  Took 11 diodes in series to prevent any diode clamp action effecting said plate current reading.  BUT if that meter opened up, their will only be 7 or so volts from ground to rf deck.

--Shane
KD6VXI

It appears that a meter in the HV- lead will read both PA and modulator plate currents on the common supply as John has.

ok on the higher voltage ratings. I actually have 20K and 30K doorknobs as well as some ruskie doorknobs that are 30K. I will update the schematic. Tell me more about the audio passthru. I have 4 500pf's and 2 470pf's to choose from.

The reactor cap is marked 3kv but it is the actual cap that came out of the gates so my guess is that is the working DC voltage not the peak. However,  that is the only component that I now have from the Gates that seems marginal in rating, so I will do as you say and add a twin. Safer approach than trusting an engineer's decision back when I was born (mid-50's). I have more risk than he does if it blows, I am sure!

On the screen, have doubled the plate relay to also turn on the screen. If HV happens to be present with the screen and plate off, what can happen?

We need to discuss the metering better. I understand all that was said but agree that putting it in the HV- will now read both plate and mod. The bigger issue is that changing the bleeder configuration breaks my compatibility with using the HV supply with my L4B. If I am going to add this I need to do it on the RF board only. Thinking.

I floated the screen and bias returns.

I have to rewire the filament supply feed from the controller. I have not looked yet but am sure the Gates fil trans will be 240V. Also, may need to buck it or add a variac to get it to exactly 10V. Going to add a filament voltage meter.

With all the relays and the variacs, I think am building Dr. Frankenstein's lab... Sparks a'flying and relays a'clacking!!!

John

Steve,
Yes, you are correct. Good observation.


I think John now has the metering all independent in his latest schematic by floating the negative for the grid and screen supplies  - and then keeping the plate current meters in the fil CT leads for the plate mod and plate RF.  That should all work FB now unless I am missing something. For each solution there seems to be a disadvantage but this is a good compromise now.


John, what do you mean by "audio passthru" caps?  Maybe you mean limiting the plate bypass and plate coupling caps to 500 pF max so that the extreme audio highs are not shunted too much.  Then yes, keep them 500 pF or so. Remember that the pi network has 1300++ pF to ground at times and that plate coupling cap is the only isolation from this being part of the mod xfmr audio circuit.  The smaller the better, as long as it works OK for RF..


Your question:   "On the screen, have doubled the plate relay to also turn on the screen. If HV happens to be present with the screen and plate off, what can happen? "

HV and plate are the same thing to me, so not sure what you mean.  But if the HV is on the tube, PTT keyed or not keyed in the fil CT, fil voltage on or off, screen off, then no problem.    The only problem is when there is screen voltage on, fil voltage on, PTT keyed, HV off.  This will cause excessive screen current and do damage.   We should probably make up a matrix table showing various voltages on or off, PTT keyed or off, fil voltage on/off to show all of the possibilities. But once you sit down and think it out, you will have a better understanding of the risks and what you should not do.


But the risk still remains... even if you have sequenced screen voltage keying, what happens if you are transmitting and the plate supply fails for whatever reason? Your screen current will soar and possibly do damage to the tubes before you are aware enough to unkey the PTT. That's why a current sensor or clamp is required in the screen circuit.  Accidents can happen to the grid circuit too but to a lessor degree.  

Once you get the rig running, you can carefully try different scenarios and see just what you can get away with and not.  IE, key the PTT, slowly add a tiny amount of screen voltage with the Variac and apply a small amount of drive... Turn on HV and see what happens, etc.

All of this is akin to driving a dragster. There is nothing automatic and safe about any of it. There's a lot of knobs and adjustments and easy to get into trouble when trying to make them all work together smoothly.

You continue to do a fantastic job putting this all together, John!!

Did I cover everything for now?

T

ok, Tom, great explanation. So screen is not an issue unless the plate disappears, right? Then the screen becomes the plate and bad stuff happens quickly. So I really don't need to key the screen. Just have a clamp on it that activates if I lose HV, right? Is it that simple?

John

John,

Yes, this is correct.

Essentially, if the HV and screen are both always on, or get sequencer turned on or get turned on together, no current will flow anywhere until the fil CT is grounded with the PTT relay contacts.

So everything will work FB until you lose the plate voltage by accident while the PTT is on.  Then the screen magnetic current sensor or screen clamp circuit takes over.  The sensor will break the PTT control line so that everything sequence cycles off.  You COULD also have a direct path from the magnetic current sensor to the screen DC circuit to break it so it occurs faster, outside of the PTT sequencer rotation.  

Or an all-electronic screen clamp shutdown circuit would be the fastest of all.

There's many options here.

T

I get it. I am planning to use the Arduino for sequencing. It can also read some current sensors and convert it from A to D. I can have set points that then override the sequencer to drop the whole shebang in an orderly fashion. This could be pretty cool.  8)

I think I may even have to find some room for a 4 Line LCD display. My DDS is already being run by an Arduino and has a Freq display.

John

Sounds slick, John!

When I have a fault in any of my six current sensors, hearing the sequencer cycle down in an orderly fashion is a thrill.  I use latched LEDs to identify which tube element caused the problem and a master reset button to try it again.


I have a question for the group related to the screen clamper, etc.  

**Update:  Never mind - question answered offline.

T

Chassis on order.

What is a good rule of thumb on how deep to make the decks to fit in a standard rack? The Motorola rack I had my eye on is just too small. Only 17in deep. I really need to make the PS and the Mod deck around 17in x 20in or so. The mod deck will have a 17x10 front chassis for the tubes and mosfet driver. It will set on a wood platform that has space in back for the mod transformer, mod reactor and reactor caps. I think I can get them to fit in 10in of space or maybe a bit more. The PS is needing a wood platform of about the same 17x20 or so. A few more inches would be even better. How about 17x24in?

Looks like the Lehman FLG globe will work for the chimney unless anyone can suggest a different one. How do you attach them to the chassis? Price is now up to $6.95, not too bad.

John

Hi John,

The boards arrived and are beautiful - thanks!

Here's the link to the "813" chimneys for the other guys. They look like the ones I used. Great price for temperature-rated glass. They can be secured to the chassis like chimneys made for 3-500Zs, etc. Some kind of clips that have spring.  Three or four clips in a circle will do.  Look around and you will find something that will work, even if they are the expensive Eimac clips. Put your blower motors on a Variac and you can set the air flow to a very quiet level but still cool FB.

If you sub-mount the 813 tube sockets by 1/4" or so, you will have a good breeze thru the chimneys. No need for air holes in the chassis.

https://www.lehmans.com/product/clear-globes-for-dietz-blizzard-lanterns

As for the cabinet depth, that really depends on what you can find. I have always been very grateful when I find a DEEP cabinet. The extra depth always gets used.  Even if not, the extra room for working and better ventilation is FB. Very few hams ever complain about having extra space in any situation... ;D  My cabinet for Dr. Love is 30" deep. Even 36" would be FB.

T

Dr. Love:  Notice the trap door when I used plug-in coils. Dr. Love now has a big bandswitch.  Look closely and notice the fiberglas air ducting for the cooling outside. That has since been chaged to a sub-chassis blower with a Variac.

My all-time favorite color combination has been a white panel, black lettering, with a "blue IBM"  cabinet.  

Below: The days of wine and roses... and sweet memories.

John,

Since you're still on the quest for a rack....

It makes a big difference in accessibility by having the rack side panels easily removeable. This includes the rear door, if any.  You can actually lean in and work on the rig when the side panels are removed.  Some racks do not have removable side panels, so be sure.

Also, be sure to have a bracket or shelf that supports the bottom of each rack panel. Most racks have brackets and/or mounting holes to start the process.  Holding the chasis up by only the 19" rack screws is NG. You will be axing the XYL for help unless you are Dick the Bruiser.

Better yet, if you are really smart, get some of those 19" rack slide in/out assemblies that lets you work on each 19" unit out in front of the rack.  Just be sure you have plenty of weight in the bottom of the rack or low enough C.G. to prevent rack tipping.

Make sure the wheels are extra heavy duty AND the wheel brackets holding them to the rack are extra strong and well reinforced. I actually had a rack NEARLY fall over in the shack when one of the front wheel brackets collapsed.  I caught it in time and had to wrestle it into a safe position. The biggest stress is when you are moving the rack over uneven or rough floor, rugs, etc..

Be sure the height fits into a room with low ceilings... or lower ceilings in the future.  (crystal ball)

Paint it to perfection as soon as possible, before you start putting stuff in. Otherwise it may never get done.

T

Me too! I wuz doing that in WallyWorld today and almost got thrown out for gawking??

always a wise guy in the crowd  ::)

I have an enclosed rack that I acquired from Joe/WB1AIU which is the home of GORT.  The rack is 4.5ft tall and is 24 inches deep. I mounted it on heavy duty wheels. Between the rack and the hardware installed in it, it's about 250lbs. It has 19" rails both in the front and in the rear.  I took advantage of of the rear rails by installing perforated angle iron between the front and rear to be used as sliders for the 4 chassis support sections of the transmitter. There are 2 chassis's with transformers and there is no way that the front panel of each chassis will support itself let alone trying to install and remove those chassis's by removing and installing the mounting screws. That would take to people.   The rails provide support and unstress the front panel and I  am able to remove/install each chassis by myself.  

Tom, after thinking about your question and because I did not arrive at any valid hypothesis on my own, I refrained from commenting prematurely.  But I (and possibly others) would  like to hear what you learned offline about the behavior of the screen grid voltage and current when using the combination of fixed and grid-leak bias as the grid current changes.  (I was out of town for a few days, hence the delay in my comment.)

Thanks in advance for taking the time to elaborate!

Hi Rick,

Actually I wasn't really satisfied with the answer I received and didn't want to waste any more time on it.  I decided the next time I fire up the 4X1 rig I will see for myself exactly how the screen current responds to grid drive as the grid leak bias increases.  It eventually settles at 280 mA with full grid drive and with the Variac-driven screen voltage adjusted right, but the path it takes to get there, I don't remember... :-)

T

Thanks for the explanation, Tom. 

Finally! I am in possession of my rack. Is an old Motorola comm rack. Built very well. Has some nice stainless louvers on the bottom that will really help air flow - and look super. Needs a good cleaning.

I have stripped off the sides and the front (now back) door. I positioned the rack bars so that the transmitter will have the faceplates flush with the now front line of the rack giving me a whopping 17x17in space to mount the mod and RF chassis. I had bought a 17x17 one but now will trade it in on a 17x15 one. Glad I was not premature on cutting on the chassis. This will be a bit tighter but will give me room for cables on the back of the chassis and still be able to close the door.

Putting in a 3/4in plywood floor and casters to allow easy movement. First task is to position all of the iron for the PS components on the wood floor. I may have to do a double decker for some parts like the caps. Gates had all of the iron on the floor of the rack. Having the sides off will make this build much easier. I am using 5/16 bolts and t-nuts to make the iron solidly attached to the unit. I have 175lbs of iron to mount. Casters can handle 600lbs of weight. I am thinking I will be in the 400lb ballpark when finished. Heavy metal baby!

I will then start on the power control chassis. Then modulator and last the RF deck on the top. Included the schematic for anyone following along.

ain't nuthin like a big homebrrrrrew project ! :D :D   like the new avatar as well

Was on the shore of Lake Erie with a microbrew in my hand. One of my better days.

A big homebrew is a bit like this game

https://en.m.wikipedia.org/wiki/Three-dimensional_chess

Just about popped a nut just moving them into position. Definitely going double decker!

Think I will put a layer of aluminum between the wood and the iron. Better ground for all the iron and no visible combustibles. Overthinking this?

John

Yes overthinking...Old Iron is actually better mounted on wood..Frames not at ground potential....IMHO

Yup, that is the way I do it too.  No need to bolt 'em down, either.  Just place them with reasonable clearance, and they will stay put, even with the rack on casters.

In my 250TH rig, I used an HP 5KW isolation transformer.  it had a slight hum, even on the wood, so i placed it upon a one-inch thick slab of styrofoam, quiet as can be now. 

By the way, when I was about five, my dad was fixing a radio for me.  It needed a new filter cap.  But when I asked him why it hums, he just said "because it doesn't know the words."

Wood is good! 3/4" is a bit thin for 'no flexing' with that heavy stuff there. Probably 2x6 or 2x10 planking is more than enough to make a great floor and will just fit side to side, depending on the rack.

Personally I like to secure the iron in place and do not understand why people like to have it loose, just set in place. To ground or not to ground? depends on the parts and safety concerns if someone else ever has to work on it.

Planks let you run 1.5" x 1/4" or 3/8" lag bolts into them to fix the iron in place yet be able to remove it easily with a 7/16" socket if wanted. Fender washers under the bolt heads keep from scratching the transformer frames.

A piece of angle iron across the planks at each end hold the 'crack' together great. I did it a little different, running transformer bolts into the angle too, but that floor is solid due to the angle iron and nothing is going to move.

Overkill? maybe, but we should each do what makes each of us happy!

Just thinking long term because this kind of equipment you are building could last 50-100 years if done right.

p.s. don't use water-resistant/pressure treated wood. The chemicals can slowly corrode metals.

I was called nuts for stating frames should be isolated.

--Shane
KD6VXI

Hola John -

After a brief interlude, you are back.  I also like the new mellow avatar.  You look like a salty dog on the sea..  ;)

Paint the rack a nice IBM computer/ blue like below..

As for big iron mounted on wood, I have always done this with mod xfmr, chokes, Heising caps, HV power transformers, etc.  The practice simply keeps an old transformer from risking it's frame to wire insulation rating.

The only downside I see is a safety risk in case of the slim chance that one of the internal wire turns touches the frame/case after a blow out and puts the frame and case at full potential. (Or the wire to frame insulation simply fails)  I've had a Heising cap, 5 KV @ 2 uFd once internally short to the case and make the case hot for HV.

Any thoughts on this?


BTW, how do you get "Two shots of whiskey" from JSW?   Is that AFTER drinking a whole bottle?

T

Whisky 9er Juliet Sierra Whisky - sound it out.

Having a hot frame on the iron worries me. I think I would want it to blow the fuses if a short developed. Look at the other side - you have a transmitter that continues to operate but you now have a risk of fire if it starts to arc due to anything bridging the frame to ground - perhaps a mouse or something. Or you fiddling where you should not and touch a frame that you THOUGHT should not be live. I think I want it grounded. Convince me otherwise...

John

Or is the idea that if you float the frame on a 60 year old transformer it will make it less likely that it will short.

Yes, that's the idea.

It's safer to ground the frame so that if the insulation fails, the HV fusing system opens and kills the HV source.

Though, there are lots of exposed HV points in the rig and if the HV power is on we have no business touching anything inside anyway.  IE, whether we touch a hot transformer case or the HV diode stack, we get zapped.

I personally prefer to float the iron on wood or Plexiglas and never touch anything inside the cabinet when the HV power is on.  Always act like the transformer case is hot. We could make up a Plexiglas enclosure to put over it, I suppose.

"Whisky 9er Juliet Sierra Whisky - sound it out."  I see two whiskeys, but how does 9er Juliet fit in?


T

If you really wanted to be safe from a shorted frame, tie a resistor to the frame of the isolated xformer / choke and use ohms law to drop it to 120 volts (or 24, depending on the trip voltage) and hit a shunt trip breaker.

That will keep the breaker from staying live under any circumstance darn near.....  Even if you lost ground connection to the chassis somewhere/how.

This is how electric is cut during a fire or other event in commercial and industrial electric.

--Shane
KD6VXI

If wood is used for a floor, then a heavy jumper can ground the transformer frame to the chassis or the protection scheme can be used. There are always options.
The pictures were only to show what I thought was a great way to make a real strong floor in a rack at almost no cost.

In my 1950's 4-1000 homebrew rig, everything is grounded to the cabinet but the mod transformer, because that item is very ratty and from the 1940s. The rest of the stuff was new at the time. The original builder did not ground the mod iron because he was sometimes running 5KV in there. When I put the thing back together I rewired the HV supplies to make only 3300V.

When designing and building something new like in this topic, there's a perfect opportunity to put the low voltage and control circuits well away from the high voltage parts in case some work has to be done.

ok, got it.

HV gets assembled this weekend.

What do you all use for a HV probe? I am thinking of buying the BK 28A one.

https://www.amazon.com/Precision-PR-28A-Voltage-Attenuation/dp/B004PA02Q8

John

That one would be great for ham radio work, just use a lower scale on the meter. I see it has a ground clip. That's the safety feature as well as the tapped multiplier return.

No matter what kind storebought or home made, I never like to hold them while measuring, but I'm shy after being hit by a color TV set because of a faulty probe.

Some of the TV set types like that 40KV unit on Amazon used to come with a slip-on spring hook that let it hang in place without being held.

I have that same probe and found it indispensable when probing HV circuits.

I used it a lot when checking voltages on the HV circuits in my 813 based 175 watt Screen Modulated, 3-band, homebrew transmitter, which had an HV B+ of 2,200 Volts.

 
Your build would make for a great article for Electric Radio magazine.  8)


Phil - AC0OB

Progress...

Have the front panel laid out to mount the power switch, breakers, HV Meter and the honk'in powerstat. Time to make some holes. Power switch is one of those NC/NO ones that works in conjunction with a 3PDT relay to have the power shut down completely if there is a mains glitch. Will also have my back door interlock in the circuit so that if the door gets opened, it all shuts down.

All the iron mounted. Tight.

Working out where to place the big caps and the relays. I think the caps will mount on a side shelf strapped to the side of the rack. A similar shelf on the opposing side to hold the 3 relays. I think that will work. 

Do the magnetics generate much heat? Mod trans and mod reactor are pretty close. Can move them an inch or so if necessary. Those are the two in the back - L3 and the white tag one.

Glad I have 6 feet of rack to work with. The plan shows it all fitting with a 3U space left over. Thinking about a 3 fan spacer and running the fans at 1/2 voltage. The bottom of the rack has stainless louvers so that should create some good airflow and hopefully not be too loud.

John

Not to pick, but laminations are facing same way on power and audio iron -but it's not a preamp being built after all!

Looks like it will all be grounded, so close-in but it's fairly new iron. Doubt you will really have any shorting or heat issues. I think it's great! Love generous power supplies.

Nice big Variac as well. Have a shot of Red Cat to celebrate!

Still fiddling with the layout.  ::)

Changed orientation of the mod trans and mod reactor to make more floor space. BTW, floor is 2 sheets of 3/4 in plywood. Plenty stout and give me an ungrounded mounting platform. Also bought some insulated terminal boots to allow me to cover up all of the connections to insure no possible arcing in the future. White wire is 20kv rated, red wire is 30kv rated. Diode string is mounted on polycarbonate with a insulating sheet below. White posts are for the HV fuse. All caps will be secured once I finalize the plan.

Option 1 fits all of the caps on the floor, but I lose one of the large caps so I go from 64uf to 32uf on the PS caps. 32uf is still 4X what the Gates had which was 8uF.

Option 2 maintains the 64uf of PS caps but I then have to place the reactor caps up high on the side of the frame. Longer leads on the caps which I generally do not like.

Which would you do and why?

John

Being pessimistic, but what about serviceability? If you mount the iron and caps like that, how do you probe anything, let alone change taps or remove one of the transformers, without taking it all apart?

Ed

Hi Ed,

With option 2, I can probe all of the taps on the mod iron. Option 1, not so much, so perhaps I will go with option 2. The taps for the PS are on the front and top and can easily be accessed by removing a louver, or reaching in the back with the door open. The next item in the rack will be a PS chassis for the bias supplies. It will be a short chassis that still gives me space to safely probe. As to removing the transformers, they are easily removed with the sides off. I used t-nuts and bolts to hold them down. Easy nutdriver access to remove the bolts.

Given that this was the only rack I could find, have to make do. It would really be nice to have 6 more inches of space. It will still be easier than changing an alternator on one of our cars.

Thanks for the comments.

John

Progress...

The meter is special. I don't recall where it came from. I have had it since I was a kid some 50 years ago, probably a ham-fest. I had it in a single meter enclosure and it has knocked around with me since college or before. I had set it up to be a 25V dc meter. I really needed to use this meter on this project...needs cleaning.

PS front panel laid out. No indicator lights as the Off-On SW lights up in the center. I need a power stat metal disc. Anyone have one laying around?

I think I finalized the iron footprint, baring any unforeseen obstacles that you guys point out. See anything odd or out of place?  ;D

John

Hi John,

Looks robust!   You are a trooper working into the hot summer when many take a building break.


Question: Maybe it's an optical illusion, but how much space is between the bare HV diode leads and the plate transformer iron?   Dust, moisture, sharp edges, corona, reactor collapsing fields, etc., can cause arcing greater than the normal 1/2" = 10KV standard.  I would have at least 1" of space there.  Or, a Plexiglas plate placed under the diode board would suffice.


T

There is a Plexiglas plate there... Your browser may allow you to zoom in to see it.

You know, a lot of guys would like to have this problem. This is some serious kick a** iron that Phil sold me.  8)

This is being built in my basement. So nice and cool down there. Kind of like building a yacht in the basement. Won't be my problem to get it out of there... My goal is to be doing trials in the fall/winter.

Has Yaz been taking you sailing this summer? Where do you usually go? Salt or fresh?

John

PS build continues. No pics, just boring stuff. Relay wiring etc. I take my time. Solder connectors, then shrink wrap the ends to make them nice looking.

One thing though - installing the power cord. Bought a 30A 4 conductor dryer cord. This is when you embrace the fact that building one of these rigs is a whole new ballgame! Last time I messed with a cord like this was wiring an arc welder.   :o

John

I would choose a layout with the oil capacitor terminals facing up in case of future leaks.

32uF vs 64uF will probably be almost unmeasurable if it's about ripple.

If the bias supply is on a smaller or narrow chassis then the caps can be mounted to a 2-3U piece of rack panel with straps, and the panel bolted horizontally to the inside of the side of the rack so the caps are vertical. I see some holes there. The power supply control panel looks like it gives plenty of room vertically if the bias suply is to be above it.

Long HV leads don't bother me except for abrasion and so I mounted some 1" PVC pipe to conduct them vertically between levels, so that the wires are not fastened against the metal. The PVC pipe is held to the rack side supports by pipe clamps top and bottom. A large one was easily bent to fit over both PVC pipes but I could have used one larger pipe save it had to fit into the exact space between 2 joined racks.

I think an 8" style variac dial scale is made for that size unit. Only a spare 4" here, too small for that!

Good on soldering your connectors. Crimping and assuming it will last forever is for the cheap seats! Nice on the double 3/4" floor.
Are you using plug-in relays or din mount or socketed panel mount? Just curious.

Have you decided on/collected panel meters for the usual parameters?
A couple rows of meters always looks great but space sometimes dictates some sharing like one meter for Modulator plate current 1/2/both, etc..

Do you plan a passive built-in 'test' multimeter to help highlight any future problem area or reduce the amount of probing or disassembly? I guess it's a luxury but it could reduce probing hot areas and save the back-ache.

Wonderful how this is coming along. Please post more pictures even of little things like relays and innovations/tricks.

 :)

Been thinking about this a lot. It seems easier than it might appear and when a relay is activated, a lamp could indicate which one, assuming it could latch and had its own fused connection to the power input separate from the power supply contactor and breaker.

How did you get that rectangular hole in the front panel for the power switch?
I have hand tools, it would be a lot of work for me to do it that way, filing and such.

Agree with all of your suggestions. Will post more pics this weekend. The bias supplies are going just above the PS on a 17x10 chassis so there will be ample room behind. I have lots of meters to mount. Just wish it was easier to cut 2 3/4 in holes. Been using a saber saw. Panels are powder-coated steel. Aluminum is way too expensive.

The switch mounts in a round hole and is pretty unique. It has a red bar that allows the front to detach from the back. So you mount it then wire it separately, then insert it and slide the red bar. It also has a 220v lamp inside that turns on when active. Got it on Amazon for $7 bucks.

https://www.amazon.com/Yohii-Yellow-Momentary-Button-Switch/dp/B07H1B94XD

Bought 25ft of 10-3/G wire to run a 220v outlet to the final location of the transmitter. Going to wire that up this weekend. I then will test the NO/NC action of the power supply up to the breakers. Bought 2 300W incancescent lamps that I will wire in series to put a load on the 220v and let me test. This circuit is supposed to allow me to turn it on and off, but if power is interrupted, it will drop the relay and shut it all down. After that test, I will attach the soft start and variac and see if I can control the brightness of the lamps. I will then be moving on to testing the 2500V HV (very, very carefully!!!   :o  :o  :o )

Started laying out the bias supply chassis also. Need suggestions on grounding. The center tap of the main transformer is where the ground is starting - ground zero. I will have this attached with 10ga thhn stranded green to the ground lug on the power cord. Do I then run a 10ga ground up the stack in a daisy chain to each of the 3 chassis? Or do I run a separate ground to each chassis from the center tap?

On the chassis do I just ground it direct and then use ground points on the terminal lug strips (using star washers) that I will be point to point wiring, or do I run distinct soldered grounds to the lugs in addition to using a grounded chassis?

John

Please don't trust star washer type grounds to make the rack serve as a ground bus, only to connect the rack to the electrical ground system whether it be a bus or a star. I have chased dozens of loose solder lugs after someone used a cabinet as a ground system via screws with star washers where the machine screw and its bolt became loose over time to where the lug could be moved by hand. If a lockwasher (cut washer) and optional additional nylock nut had been used it would not have happened, but still, a steel cabinet makes a better shield & safety device than a ground, and the good connections offered by a copper wire make a better ground system.
Nylock nuts are costly when used everywhere. I only use them to secure the mains and B- ground to the rack, and the blower motor frame since the blower said it has to be grounded. They are a luxury, not needed when a cut washer is used to maintain tension on a stack of stuff on a bolt but I like the extra peace of mind.

There are are many schools of thought on grounds. The bus and the star are most simple and good choices. Each has good uses. I'm no expert on this topic and the next 3 paragraphs are my opinion which may have many flaws, maybe someone who knows more will kindly comment.

The bus is good for power circuits where a few mV of voltage caused by varying currents in the bus itself won't interference with any of the units using that bus for a ground.
When it is used for a DC and signal return both, then the main ground, which is the mains ground + power supply B- ground + the chassis ground + the signal ground of the weakest signal stage are all tied to a common point, and the bus runs away from that point to serve the circuits having higher and higher currents.

The star is best when there is a variety of large-and-small signal circuits to be grounded. Depending on what is to be rejected, the ground point might be the signal ground or the mains ground, or both. The point at the middle of the star has to be the best ground possible.

Good bypassing or a line filter on the DC power input to a chassis or the mains power input to a chassis is usually enough to prevent RF and high frequency noise from entering the chassis' and that should make either kind of bus acceptable because things like oscillators should have their own regulated power supplies not affected by mains noise or its relation to mains ground or chassis ground. Although the chassis are 'grounded' by means of being bolted in, each chassis should have its B- (C+ in case of bias supply) coming to a terminal on its terminal block. (that is also usually connected under-chassis to the chassis itself, to serve as the mains or safety ground).

Solid is best for stiffness of a bus but stranded takes solder much better.  The two pics show bus and star. There are other more complicated ways but I think your choice of these should be worry-free. The 'bus' picture would easily be the daisy chain you mentioned. The ground bus would daisy chain from one unit to the next.

Patrick,

Think I will borrow this design to protect my screens... I have an unused 6.3v winding on the transformer next to the screen supply. Nice straightforward design.

http://amfone.net/Amforum/index.php?topic=43765.0

John

It works great too. Have tripped the screen quite a bit in experiments -totally reliable. The circuit was built into the transmitter By COL Tucker when I got it and includes a lamp and pushbutton reset on the front panel. I merely drew the schematic of the beast.

Relay seems to be made of unobtanium. Any pointers to one or a modern equivalent would be helpful.

Guardian X-300-ER overload relay

This relay is so simple and foolproof I really want to try it, but I have a backup plan.

ACS712 current sensor in the negative line of the 400V supply. I will sent the analog signal to the arduino sequencer. While in PTT mode, I will loop on sensing the voltage on an analog input. If the voltage goes too high i.e. too much current, I blow out of the loop and activate the PTT off sequence. I don't think I need to shut off the screen drive with a specific overload relay, just run the full sequencer to off should do it. Seems simple enough. Need to think about it some more. I do have a relay on that line in the schematic and I have extra sequencer capacity so I could preemptively kill the screen then blow the subroutine loop.

Wonder what the MTBF of the relay is WRT the MTBF of the software/hardware stack.  ???

John

MTBF on most relays is 100K operations. Mine is almost 70 years old.
The arduino 100-year paradigm -who can say? Glue a heatsink on the processor, underclock it, add a fan, shield it very well.. No way I would know. Most semiconductors have a life expectancy of at least 20 years according to a previous employer who manufactured them. There might also be a high-rel version. The other parts on the board may have life expectancies too. Keep it cool and free of electrical abuse, right?

Mechanical (dual coil) latching relays:
Mechanical latching relays like the old-style one mentioned occasionally show up on ebay. If I had a spare I would give it.
I guess MTBF, if used to only control the small coil of a power relay, is 100K operations.
Mouser selection tool:
https://www.mouser.com/Electromechanical/Relays/_/N-5g31?P=1z0x216Z1yzxc13Z1z0x3udZ1z0x3bdZ1z0x3tdZ1z0x3txZ1z0x3w0Z1z0x205Z1z0x3ubZ1z0x35nZ1z0x3vzZ1z0x3ylZ1z0vo2rZ1z0x3brZ1z0x1diZ1z0x36rZ1z0spixZ1z0vkvbZ1z0z1s3

Magnetic latching relays:
Some modern latching relays use a magnet to assist and are supposed to therfore save energy.
They latch and reset by the user reversing the polarity of a momentary voltage applied to the coil.
They can suffer from magnet weakening over time.
Mouser selection tool:
https://www.mouser.com/Electromechanical/Relays/_/N-5g31?P=1z0z1wpZ1z0x216Z1z0x3bdZ1yzxc13Z1z0x3udZ1z0x3tdZ1z0x3txZ1z0x3w0Z1z0x205Z1z0x3ubZ1z0x35nZ1z0x3vzZ1z0vo2rZ1z0x3ylZ1z0x3brZ1z0x1diZ1z0x36rZ1z0spixZ1z0vkvb&Ns=Coil+Voltage%7c1


I guess the spec. is how much voltage & current is needed to unlatch the relay. Usually they are kept latched in normal operation since it takes more juice to latch than unlatch, at least on the old one I have.

By no means am I innocent of putting electronic relays in my equipment. The AMF PB time delay units have a board with transistors governed by RC circuits.

Patrick,

After looking at your schematic more closely, I see that it needs an ac 6V coil to latch. Does it also have a AC coil for the reset?

Galco RR2KP-U-AC6 - Latching Relay, DPDT, 10A, 6VAC Coil

This company seems to be the successor to Guardian. I see how to use this relay for the latch of the transformer primary. Not sure if if will work on the negative side using a WW pot with an AC coil.

I have 24VDC available so I could use a 24V DC rated coil pair. 24V to latch for operation. Then set the trip at 150ma. That equates to a 190 ohm R in parallel with the reset coil to give a 19.2 voltage drop that will trip the reset at it's 80% value. At 80ma, the drop is 10V which is not enough to trip the coil. A 400 ohm 5W WW pot should give me plenty of capacity.

Am I on the right track?

John

Continuing to wire the HV today. Thinking about spark gaps.

Attached is the preliminary thoughts. The top of the big iron have flanges that is perfect for attaching lexan sheeting that I can place HV parts on. I did this with the diode strings. I thought about a common metal AL strip grounded to the HV center tap. I would place 3 spark gaps along this strip, each having a metal AL strip that has an adjustable elongated hole. That way I can adjust the gaps appropriately. I may then make a clear LEXAN box to cover it.

I plan to place gaps at the CT of the MOD primary (2), the Plate choke side of the MOD secondary (5), and the junction of the heising choke where it connects to the caps (4).

Is this a good plan and enough protection for the mod iron?

John

The reset, or 'trip' coil for the screen current protection action is DC type appropriate for DC screen current. I don't think I have the voltage/current specs on it or don't remember. I am sure the AC 6.3V latch coil was mainly for convenience when 6.3VAC was more common.

If mechanical latching relays are what you want then I think you are totally on the right track there. Those relays are really appropriate-looking as well. If you have DC power then it's perfect since the screen current is direct.

you can sumulate mechanical latching relays rather easily if you will integrate needed funtions into the hv start / stop relay (regular non-latching type):

                                                       HV on                                       HV off      trip relay
                                                         ...|...
                       .................................o   o...............{)()()().............o..|...o.....o....>o.........
                       |               |                   no        |          coil                    nc            nc            |
                     pwr             |          no                 |                                                                 |  
                     supply         ........o        o.............                                                                  |
                       |                         \    relay                                                                             |
                       |                          \   seal in                                                                           |
                       |...|


this is crude and simplified but shud get the idee across .... without the seal in contact the hv relay only remains on while the hv on button is pushed.... as many trip elements as you need can be used by adding normally closed contacts in series

(please connect the bottom common line across the page ...somphin in the editor won't let me do this  ??? ??? ???

slight rearrangement due to LTspice symbols on hand, but includes the pesky wire.
240VAC coils all, except trip relay which is to be chosen as desired.
suggest magnetic breakers instead of fuses. cost more to put in but trip as fast as fuses blow and cost nothing to reset.

You guys are going to have to break this down for me a bit.

Here is what I thought I was doing - running along nicely until the HV supply blows a HV fuse (not the breakers). Immediately thereafter the screen current goes quickly above nominal. This should trip the overload relay and shut down the 400V screen thus saving the tubes. Perhaps I should have my sequencer see this event and then shut down everything else or maybe I see it go dead and release the PTT manually.

Why do I want to add trip functionality to the 240V input to the HV PS? The HV is already off.

Please explain...not understanding.

thanks,
John

  My Collins KW-1,   primary of 500 VDC screen power supply is hooked to the primary of the
2,400 VDC plate supply. If the plate supply goes down, the screen goes down also. There is a
 overload trip relay, in series with negative high voltage to final tubes which has a set of contacts that the screen voltage thru, so if it trips out, screen voltage disconnected.

    In the control panel portion, there is a tune, operate switch, In tune , the 500vdc, is disconnect from the wiring to the rf screens and the rf screens are chassis grounded. In operate, the screen voltage wiring direct to the tubes.

   This is nice for tuning for rf resonance, you don't have cherry red tubes, plate meter pegged
while you trying to dip and load the amp .I can dip , load to 150 or so milliamps, switch to
operate, screen voltage on tubes, 300 + plate current.
 
  The AC circuit, fused, both sides  thru 25 amp  motor starting relay controlled by a HV
standby and then manual or ptt transmit

Bob  W8LXJ

Hi Bob,

Sounds like a cool setup. The screen primary winding connected to the HV primary is a good idea.


I don't have the schematic, but what would happen if the HV secondary winding opened up or the HV diodes open up so there was no HV on the tube plate, but there was still AC primary voltage - so that the screen voltage was still on the tube?

Would the negative lead relay cover this or not?  IE, if HV is not there, does that relay drop out?

John: I open the PTT keying circuit whenever there is an overload fault. With tubes, usually the voltages will bleed off and not cause any problems. Solid state stuff will not have the same sense of humor, however.

I use Hall effect devices in series with the negative DC leads of the plates, grids and screens of both the finals and modulators. They are preset to trip on current overloads and open the PTT circuit automatically when faults occur. There are cheap Chinese Hall effect devices on eBay that are easy to set up and trigger a latching multi-relay board also found on eBay.


To answer an older question:  Yaz IV,  is my 2 1/2 year old English Springer Spaniel and he goes sailing and kayaking with me on the freshwater lakes and rivers here in CT. I took Yaz III out on the rough ocean in RI a few years ago and it was quite sporting, but he loved it... :-)

Your plans are coming along very well there. I see you decided to use the pi-network grid input circuit with no neutralization. With a tetrode, it will work very well - no worries if you build it like Chuck's layout.

T

yes, you can do exactly that. The transmitter here will key up with the screen overload tripped. With the expected results of little to no output.

Sometimes another failure will cause excessive screen curent, like a tuning/loading issue or some fail in the HV reaching the plate of the protected tube, while the HV supply is still on and otherwise running. Some component failure like a HV fuse blowing or plate choke opening or even bumping the loading control could cause the screen current to go up.

The only thing the protection has to really do is stop the excessive screen current and there are so many choices on how to do it.

So, if the main concern is loss of plate voltage causing screen current issues, wouldn’t the simplest solution be a voltage divider and a relay/sense line on the plate voltage to enable the screen supply? No unobtainium overload relay needed. You’re already using a fancy sequencer thingy right? Just another input to consider, include a brief wait on power-up only to let the plate supply caps charge or the screen won’t ever enable, and code it so the screen enable is latched off once tripped.

It seems to me sometimes, in our quest for perfect, things get way too complicated. Ive started weighing the cost of the protection and time to implement it, vs the cost of the parts being protected.....I want to make RF and get on the air...someday...

Ed

The problem with sensing the screen trip off the high voltage is WHERE do to you sense the HV?  If it is anywhere but at the plate cap of the tube, then we could have a failure.  IE, there are many places the HV supply and HV circuit could fail - anywhere from the 240 AC up to the plate cap. And what if the factory plate cap weld goes intermittent sitting on the tube glass? I've had that happen too...  Plus, there's RF to contend with at any point past the bottom of the plate choke.

The best way to protect the screen is with a current sensor in series with the screen circuit or clamp tube, etc.   There are many ways but my ultimate method is using a tiny Hall effect device that senses the magnetic field thru a wire and can then trip off the screen in any way you desire. Then the HV can do anything it wants independent of the screen.

As I mentioned before, I use six Hall effect devices into a multi-latching relay board for the grid/screen/ plate circuits of BOTH the modulators and finals. This way it becomes "uniform" for all devices with the same simple circuit employed everywhere ... easy to use and easy to remember one circuit design when troubleshooting the whole protection system.

T

Hi Tom,

You have convinced me - again.

I plan to use the same unit you used - ACS712, I believe. It puts out an analog signal with it at VCC/2 for zero current. How did you hook this up to the latching relay? Did you use a comparator circuit like is described in the application note that allows you to tune the current trip points? Since I am using a Arduino for my sequencer, I am really tempted to use one of the analog inputs and monitor the voltage directly. I can do this in a loop while I am in PTT mode. If I see any of the currents go past normal, I can shut off PTT directly. I have a small TFT display so I plan to also display the data digitally.

My only concern is what happens if the Arduino messes up. Using the Arduino is an untested approach as I don't see anyone else doing it. If RF scrambles it's brain I may lose the whole rig. So I am tempted to also have a failsafe discrete system that watches the same sensors and drags PTT low outside of the sequencer.

I just don't know...

On Spark gaps. I was planning on gaps on the mod secondary tap 4, the other side of the mod reactor secondary tap 5 and the 2 primary connections that go to the mod tube plates taps 1 and 3. Does that sound correct?

John

Hi John,

If you put everything in an aluminum box and then bypass the input leads with .001's (and toroids if needed) , there should be no RF in the circuit.

This six relay latching board below for $9.37 should work FB.

https://www.ebay.com/itm/6-channel-12V-latching-relay-module-Switch-controls-the-high-voltage-H-current/121951095953?hash=item1c64da0c91:g:oVYAAOSwAvJXCcQQ


It will take a low level signal to trigger.


A sensor like you suggested  below may work OK -  or some other derivation. These are only $3.95 so worth the try.  Order the six relay board and then see what sensor works best with it. Play around.

https://www.ebay.com/itm/30A-range-Current-Sensor-Module-ACS712-Module-Arduino-module-NEW/382824575626?_trkparms=aid%3D555018%26algo%3DPL.SIM%26ao%3D1%26asc%3D57476%26meid%3Dec2b38a14a9541fd80c52ea43f41aaff%26pid%3D100005%26rk%3D1%26rkt%3D12%26sd%3D202681517336%26itm%3D382824575626%26pg%3D2047675&_trksid=p2047675.c100005.m1851

You can generally use a 0.5 to 2 ohm resistor in the negative DC lead for the grid, screen and plate to get a tiny DC current sample.  A pot across the resistor sampler can then adjust the trigger point. At least that's what I did here.  My particular current sampler has an opticoupler in it, but I cannot determine what kind it is due to no labeling of value on it.

I did not use a comparator, but that would work fine.

This really is a cool system once working. For example, I can tune my 4-1000A finals off resonace, drive it for too much grid current or load the final for more screen current and off goes the PTT. Even a bad antenna, forgetting to turn on the HV, etc etc, will trip it. Not to mention when I mess up and hit the audio too hard, the PTT drops out due to the modulator current(s) (grid, screen or plate).  I realize there is a slight delay due to using a relay and the PTT unlatching, but with tubes it is plenty fast enough without causing damage.  When the rig has been pretuned and turned on again, you don't even know the protecton circuit is there other than the trip lights are off.  When it trips, I know exactly what circuit (of six) tripped by the LED being lit. Just reset it and try again or adjust the current sensor a little until the rig runs flawlessly at normal parameters.

T

John,

I think you have it, but just to be sure:    Four sets of gaps are used. Round ball gaps are best.  One set across pins 4-5 secondary.    One set across pins 3-2  primary.    One set across  pins 2-1 primary.    One set across the modulation reactor, L1.    I also put a small one across my screen choke as a precaution after having unkeying spark problems in general, but is probably not needed there.

Also remember that later on in testing if you have unkey sparking, a vacuum relay that puts a 5K 100 watt resistor across the mod reactor when unkeyed will quench the sparks. It solved all my sparking problems despite having a sequencing system.   No matter how we look at it, that energy stored in the mod choke has to go somewhere during shutdown and if it doesn't  eventually find its way out to the antenna in the form of RF, it will spark instead at the reactor. Bad sparking is what damages our iron in the form of internal insulation arc-overs.. (both mod xfmr and choke)   Normally commercial BC iron is safely turned on once a day or less. Hams beat the crap out of it with the rapid PTT. We gotta have it set up right or get your wallet out again.

T

Ian's (G3SEK) has a tetrode protection board that does exactly what you're looking for.

Schematics are freely available for the googling.  Would give ideas in pictures on what is being discussed.

--Shane
KD6VXI

What about using either 1 or two of these to regulate the grid current to 80ma - IXCP10M90S

An interesting thought... using a solid state current regulator device to limit the maximum current thru the screen to 80 mA. It should be invisible until current hits >80 mA.

I've never tried a current regulator so not sure if it makes sense in this application.  Maybe one of the guys with circuit modeling software can simulate it. Make sure it will handle 500V X2 if it is referenced to ground...

T

Here is the article - http://amfone.net/Amforum/index.php?topic=30947.0

I am also looking closely at the tetrode board. It has a lot of really nice features...

Tom, were you thinking of something like this?

R3 senses the Isg current and develops a voltage drop which switches Q1 on and startes to pull down the gate voltage. One trims R3 for the threshold sensitivity and current limit.


Phil - ACC0OB

thanks Phil. I will breadboard this soon. Since it is very close to the regulator you designed for the mosfet board, I am going to take one of those PCBs to build it. With a few jumpers and some extra holes for the 2N3904 I think it will fit up nice.

I think I will scrap the 75V bias design using the VR tube as well and build up another regulator board for the -75V.

Thinking about all of the PCBs and silicon on this transmitter I think I have an idea for what I will call it. With these 2, the 2 for the mosfet board, the mosfet board itself, the RF driver that will be silicon based, the sequencer that will be an arduino with relay board -  it is adding up.

The name is The Borg...

If you are familiar with the Star Trek canon, this makes sense. Take technology from other species and craft it into a new being that is technologically superior to the prior beings. With the amount of silicon crafted on to a hollow state design it kind of makes sense, I hope...  ;D

John

"     If you are familiar with the Star Trek canon  ... "
   

I worry about these 'hybrids'. You never know what they'll do.....   As Capt. Kirk would say, " whose your daddy, Nomad? "

klc

Or, how about "GLASSFET?"   ;D  The Borg is not only exhilarating but also assimilating! 8)


Phil

Still plugging along. Mostly mechanical work...

First chassis is in place - power control. Will start mounting low voltage stuff as soon as the HV is done. Probably have around 2 hours more on the HV to complete it. Still spending a lot of time outside and not in the shack. Hopefully this will start to move more once the nasty weather starts.

John

John, I see the picture, and the chassis. Have you considered using stuff like this for chassis instead? To me, while it is a commercial product, it is neater and screws right to the rack rails like it should.

(https://www.middleatlantic.com/-/media/middleatlantic/images/products/accessories/rack%20accessories/01iso/ch2-iso.ashx?h=350&w=350&bc=FFFFFF)
https://www.middleatlantic.com/products/accessories/rackmount-storage/chassis/ch2.aspx (https://www.middleatlantic.com/products/accessories/rackmount-storage/chassis/ch2.aspx)



(https://www.middleatlantic.com/-/media/middleatlantic/images/products/accessories/rack%20accessories/01iso/ch3-iso.ashx?h=350&w=350&bc=FFFFFF)
https://www.middleatlantic.com/products/accessories/rackmount-storage/chassis/ch3.aspx (https://www.middleatlantic.com/products/accessories/rackmount-storage/chassis/ch3.aspx)



Big giant caveat:  my company is a dealer for this product. But I am happy to sell to anyone with a callsign for 5% over cost, plus shipping. And, I have been using them in my own projects for years -- these and their economy models (all steel, not as deep).

Ed

Note to mods: if this is inappropriate, please delete and let me know.

Thanks for the info, but I like the direction I am already taking. Plus I already bought what I need.

Looking good, John.

Suggestion: mount your power resistor(s) vertically for better convection cooling.

Thanks. Been trying to decide where to put them.

Good use of the sides of the rack though.

ok, hope to fire up the HV tomorrow. Please offer advice so that I do not fail to cover an aspect of this bringup. This is the plan of attack.

I have a 4 wire panel using a ganged breaker feeding a properly grounded 4 wire cable into the frame.

I plan to bring this up in stages.  

The old iron I will be using is a Gates transformer and matching swinging choke, into 64mfd of oil filled caps. The only load will be the 100K bleeder. I will be using a 24V separate supply for the control voltage to the main HV relay2.

I will have a validated HV meter across the bleeder that is connected such that I don't have to probe it. I will not be touching anything in or near the frame while under power. The HV will not be connected to anything beyond the bleeder at this point.

Referring to the schematic - I plan to first disconnect the output side of the circuit breakers and connect the VOM across those open contacts to monitor 240V.  I should then be able to test the RLY1 fail-safe part of the circuit to insure that it actuates the relay when on then off. Also will cycle the breaker to see if it drops but does not restart the power.

Then I will shut off the breaker and unplug the unit. I will reconnect the output 240v from the CB to the Relay2. Will set the rheostat to minimum. Will plug the unit back in and activate the breaker. Will press the ON with a wooden stick. Hopefully no smoke and I will hear nothing. Then actuate the 24V HV control. I should hear a click of that relay, then shortly thereafter a click of the softstart RELAY9 and see some indication of HV on the external meter. Since my internal meter will not be calibrated, hoping for some deflection there.

If I hear no softstart, shut down the main power and the breaker. Then advance the rheostat to close to mid point. Try again. At some point I should see some HV. Alternatively, I could install the Rheostat bypass and bypass that part to prove the softstart works with normal 240V.

Now what? The trans and choke are insulated on the mounting platform board. How do I check for leakage to the frame safely?

At some point, I should be able to convince myself that all is working, and I should be able to then safely calibrate the internal HV meter. I will do this very methodically such that I am sure there is no HV in the rack when I reach in to turn the pot on the back of the meter.

Hopefully this all goes well and we get a solid 2500V out of the bringup.

John

John, if you have the means to do so, sub in an undersized/smaller main breaker while you do this...something that will blow if something’s wrong before enough current flows to break parts.

Also consider hanging some lamps on various points to verify you have 240 and 24 as you power up.

Ed Walters

This is a 30A circuit. I was planing to use a 20A to start. Should I go to 15A?

24V is coming from a cheap switcher supply that I purchased and has it's own power LED. I will be able to test the actuation of the HV relay with everything else off.

The main switch has a built-in 240V light that is wired after the CB. Not shown on the schematic. I will place a second VOM across the trans primary to see that 120-240 voltage. Don't have any 240V lights. Good suggestion.

Thanks for the ideas.

John

John,

You could follow Phill KA4KOE's example of when he powered up his BC1-T for the first time.

https://youtu.be/fpnKwQVJN5A

Having someone away from close proximity to the rig with access to the power plug would be a good idea.

I remember once when I had an AM KW, one time the PTT relay fused together while the plate current meter was pegged, and final tubes putting an orange glow on the ceiling. I had to run through the house, into the garage, and flip the breaker! Problem was, I had push pull cross neutralized triodes. The neutralizing caps at the feedthrough null were too close together. I got this really bad idea to add 3" diameter Weston meter glass between the neutralizing cap plates. The setup null with just drive was perfect! When I keyed the B+, I darn near burned the house down...The glass shattered into a million pieces.  :'(

Jim
Wd5JKO

Watching the video - a fire extinguisher seems like a good idea!

On the floating iron - should I ground the frames of the trans and choke and only float the modulator and mod reactor?

For this test, I think I will ground the frames. That way if there is a winding failure it should take out the breaker hopefully without a lot of smoke?

If by 4-wire breaker you mean you will break the electrical ground, I believe that is unsafe.
How about keeping a solid ground from the mains GND to the chassis and just breaking L1-N-L2?

A 4 wire panel with separate G and N. Breaker cuts off L1 and L2.

Was going to do more wiring but the Son-in-Law needed the shrink wrap heat gun, so decided to hack out the sequencer code. Pretty simple code. All it does now is sequence the relays in a forward manner for TX then reverse for RX. Each step has an independent time interval. I really like the TFT display, although I am just using it for simple character output now.

I plan to put in current sensors and temperature detectors. Current sensors on key circuits and temp sensors just for grins... Will place them in the wait loop while in transmit so I can have it pull the plug if something pops.

Yea, I don't actually have the relay board hooked up. The display covers some of the key pins for power and ground, so I had to order a shield board to be able to wire it properly. Yet another item to buy. Good thing these arduino parts are cheap.

John

What is your parts count so far?

Confidential

Started working with the Mod deck. Doing initial layout. 10x17x3 chassis. Have to lay out 2 tubes, 1 10V/10A trans, Mosfet driver board and 2 100x100mm heatsink mounted regulators. Got it mostly set out, just going slow before I commit to making holes.

Tubes will mount under the chassis about 1-1.5in deep. Big black heatsink is the driver. Board on the back. The al heasinks are there for spacing. The ones that are on the way will mount the same except a 280V regulator board will be attached to the side and they mount vertically. Trans for that supply will be underneath (toroid trans).

Struggling with what to do about the filament choke. Leads on the 10V 10A transformer are short, around 4 in long. I could get a ferrite rod and bifilar wind some insulated #12 wire from some romex. Could maybe try Tom's ferrite tube idea if I had a better picture in my mind on how it is accomplished.

Wish I had some large punches. Will use my largest then the nibbler to get the tube holes cut. Priced 2.25 and 2.75 punches on the bay - way too much $$. Labor is free.

Meters - One 500ma for Mod current, one 10V for filament voltage. I have a variac if needed if fil voltage is not right.

Need some ideas on portals for the tubes. What shape of hole or holes and how to cover the holes. Maybe 2 racetrack oval shapes?

Decisions, decisions....

This deck will go quick, then on to the RF deck. I decided to wait on testing the HV. If I kill myself then all the later fun will be missed.  ::)  Decided to keep building then do the testing when relatively complete.

John

I am wondering why you need a filament choke on the modulator.  If it is common cathode, grid driven, just bypass the filament pins to ground at the sockets.

On the RF deck, I assume you are also using grounded cathode, grid driven configuration.  This also needs no filament choke.  Grounded grid needs a filament choke in the RF deck, but if you are plate modulating, you really do not want grounded grid. It will not be possible to obtain 100 percent modulation, due to feed-thru power, unless you also modulate the driver stage.  Keep it simple, grounded cathode and plate modulate the final.  You will also need to modulate the screen of the tetrode to obtain linear modulation.

For the peek-a-boo holes, I prefer the rectangular holes, relieved in the corners by a circular, rather than a square cutout.  This is very easy to achieve by drilling the four corner holes, then cutting a rectangle via a straight line on a tangent with the outer side of each hole.  Look at my 250-TH rig for an example.  This shape looks very nice with 813s as well.

I have found a hole saw works very well for the larger tube sockets. Exact size is not critical.  One way to stabilize the work is to mount a piece of wood to the bottom side of the chassis top after some of the other holes are made.  Then you can easily clamp the wood on a drill press table.  Cut slowly and you will get a very clean hole.

I am not sure that it is wise to sub-mount the 813 bases.  Not needed for shielding.  It may also restrict natural convection cooling of the base pins.  Fan should not be needed for 813s.  I prefer surface mounting.  Make sure your panel is high enough for proper plate connector clearance

Thanks for clearing that up Rick. Got confused on the filament choke usage. On with the build!

QUOTE w8khk.....
For the peek-a-boo holes, I prefer the rectangular holes, relieved in the corners by a circular, rather than a square cutout.  This is very easy to achieve by drilling the four corner holes, then cutting a rectangle via a straight line on a tangent with the outer side of each hole. 

..........Peek a boo holes in my PP 100th X 813's Rig.......

Now that is some fancy artwork!  I like it!  Could you do that for my P-P 250-TH rig?   Hi Hi

Pat Bunn N4LTA generously gave me a vfo board that he is using for his class D project. I thought it would be nice for this one. Gave me the chips as well. Got it built and tested.

Also, laying out the RF Deck.

John

What size punches do you need?

2-3/4 in for meter holes. I found a 2.25 one for $25 on the bay so I bought it. Will make the tube holes and peek-a-boo holes much easier.

Unfortunately, my set only goes up to 2 7/16” so I can’t help you.

I appreciate the offer, though..

Gorgeous rig!

I think that I have a 2 3/4" Greenlee punch if you want to borrow it. It is big and heavy.

Pat
N4LTA

Email sent!

Have a 5 day break so hope to make considerable progress... Going to try to complete the Mod Deck and then play more with the layout on the RF deck. Trying to keep it nice and tight so that it is capable of going to 20M if I want to later.  

N4LTA has a punch on the way so all of the meters will get mounted shortly, as well. Thanks Pat!

A few items to discuss -

Parasitic Plate Resistors. The schematic shows 50 ohms at 5W. I can find ceramic 2W 50 ohm easy for a buck or so each. Having a devil of a time finding non-inductive 5W units. Will 2W be sufficient? If I use 220ma and 50 ohms, I get 2.42 watts so I am guessing not, but I don't know if estimating that the current at the parasitic frequency will be 220ma. It has a DC short across it. Won't the parasitic, if present be at a lower current than that? Confused.

I will have the tank assembled this weekend, hopefully. Got the band switch rebuilt and it now gives me consistent readings on my LC meter. Whoever had it prior had done some jumpering for 160M. I want to check it out at frequency to see if I need to remove some of the jumpers. I want to play with the MFJ-259 testing tank resonance. So I need to calculate the impedance of the plates - I estimate that it will be V / (I * 2). K is 2 for Class C. V will be 2-2.5kv. What to use for I? The tube spec shows 220ma per tube at an ICAS voltage of 2250. So do I use 440ma or something a bit larger. Is 500ma a good estimate? That would give me a value of  2500 ohms. Is that correct?

I have a UTC GC-124 filament trans for the RF deck. I cannot find a pinout diagram. Reading the specs it can be run at 105, 115, 210, 220 and 230. I have 248V or 124V. I am planning to use 124V because that is the variac I have. Using the VOM I see that 1-3 is independent of the rest, likewise with 4-6 and 7-9. 7-9 has lowest DC resistance, 1-3, 4-6 are identical. So I surmise it has 6 input pins (1-3, 4-6) and 3 output pins (7-9). So I am asserting that 1-3 is a 105-115 input and same with 4-6. For 115V I use 1,2. 1,2 has a higher resistance than 2,3 so I think the picture is correct. For 105V I need more primary so it would be 1-3, right?  If I feed 105-115V I would parallel the  primaries? For 105V use 1,4 and 3,6 right? For 115V would I parallel 1,2 and 4,5? And I will still have to use a variac to keep it at 10V under load. Hope I explained this well enough. See attached.

John

You’ve got it backwards on the primaries.  Think about turns ratio and voltage ratio.  If you had say for easy math taps for 110 and 100 on the primary and a ten volt secondary, the turns ratios would be 11:1 and 10:1 respectively.  You need the more turns connection, the full winding, for the higher voltage connection. This is why the secondaries are in series for the 220/230/240 type connections, more turns=higher ratio, 20:1 vs 10:1.

Ed

Thanks, Ed! Wire it this way, then?

John

Yup. Exactly.

Ed

On the supressors, I can go with 2W x 2 at 100 ohms in parallel with Ohmite OX/OY series. Ceramic.

Or I found 51 ohm Metal Film at 5W. https://www.digikey.com/product-detail/en/ROX5SSJ51R/A142769CT-ND/10231883/?itemSeq=310514103

Ohmite WNE series (non-inductive wirewound) has 5W values but gaps from 15 ohms to 100 ohms, no 50 ohm value.

So, Ceramic or Metal Film?

Hi John,

Look for the "Glo-Bar" style power resistors for plate parasitic use..  They are sold under various names. Used, they are about $5- $10 each for a 50 ohm, 10-30 watt if you can find them.

Those wire wound 'canceling" resistors are NG for RF. Whatever you use, put it across an MFJ analyzer and sweep it up to 28 MHZ and see how it does.  The metal film type are usually OK, but a Glo-bar is best.

They are expensive but are the best. I always buy them at flea mkts when I see them.

https://www.google.com/search?hl=en&biw=1106&bih=951&tbm=isch&sa=1&ei=cdPeXaKICaTl_QaBroK4AQ&q=glo-+bar+resistors&oq=glo-+bar+resistors&gs_l=img.12...21907.22412..25073...0.0..0.99.317.4......0....1..gws-wiz-img.SqW-c0Ul1WM&ved=0ahUKEwjihOXulIvmAhWkct8KHQGXABcQ4dUDCAY

T

Some thoughts on those resistors...

From a dc or audio (modulated dc) perspective, they’re not there. The coil around them is essentially a short circuit at those frequencies.

In the 80m range, ten turns 3/8 diameter, 3/4 long (typical 2w carbon resistor sized) is 10 ohms of reactance...so the coil is still dominant and the resistor is carrying much less current than the coil.

At 60 MHz (VHF, as in suppression) the coil is a hundred or so ohms reactance and now the resistor is in charge. It’s role is to decouple, to reduce system gain, until there’s not enough to support oscillation.

There are many amplifiers and transmitters that don’t use resistors at all, they just have the plate lead coiled up, usually made from a piece of strap rather than wire.

I’m going to claim (I’m sure there are some here who will disagree) that the resistor is a coil form, and it should be seen that way. If it’s a parasitic suppressor, it’s only going to have significant current/power dissipation if the amp is oscillating at VHF, and if that’s so, you’ve got other, much bigger problems, than if that resistor is 5w or 2w.

Lastly, a note about metal or other film resistors: when manufactured, their value is trimmed by laser cutting out part of the film.  In an axial, leaded resistor, this trim is usually a spiral slot in the conductive film. A spiral slot makes it a coil. Film resistors can have some inductance, greater than that of a bulk resistor (is carbon composition, etc.) which may or may not matter in a given application.

Ed

Modulator progress. Big black heatsink and board is the Audio driver. The 2 smaller heatsinks and boards are 280V FET regulators. They are fed from a toroid transformer mounted underneath.

Think I will mount the filament transformer lower in the frame. No real good place to put it topside and I fear that it will induce hum if it is too close to the audio boards or supplies. Another octal relay will fill the big hole.

I will cut the peek-a-boo holes and the meter holes tomorrow. Then will do some final metal work and it will be ready for wiring.

Do you guys cover the peek-a-boos with any tempered glass or something?

John

J,

Lookin nice.


klc

John,

   I see the Borosil glass lamp chimney's. I looked them up,

http://www.borosil.com/products/lighting/

  Just wondering, as implemented, how do you move air through them?

Jim
Wd5JKO

I got tired of waiting for Lehman's to get more stock of the lantern globes so went searching. These popped up. I mounted the sockets under the chassis with a 8-32 nut to make a 1/8 in space between the socket and the bottom of the chassis. I cut the hole 2.25in which gives a 1/8 in space around the base of the tube. I mounted a small 12v fan that I will run at 9V so keep it quiet. I will fully cover the base of the chassis so the air only has one place to go.

While these lanterns are only 1/2 high I figured it will still properly cool the base of the tube. I will be running the tube at 2000-2500 volts. So I am hoping this give it adequate cooling. These tubes are advertised as not needing cooling under normal conditions. I am near/at the max ICAS rating so a little bit of cooling should be enough, I hope.

I got the lantern glass at a company called American Mantle. They were $7.50 plus shipping. I ordered 4.

John

Very nice, John.  Professional look.

I would suggest some improvements: Remount the 813s  with 1/4" socket sub-mounting -  and drilling 3/16" holes around the rim (between the mounting screws) to add air flow. (Drill 16 holes total, 4 per section between screws)  This will make a huge difference.  One small 12V fan on the back of the RF or Mod chassis will hardly do the trick the way it is.

If the 12 V small fan doesn't do it, try a small squirrel cage blower that can handle some back pressure using a Variac to slow it down. You should feel a moderate breeze coming thru the chimneys.

In addition, usually the chimney goes all the way to the top of the tube and bends inward to push air past the plate seal.  You will find no air hitting the plate cap with a shorty chimney like that.  I wonder how much difference it would make if you placed another chimney on top of the existing one?  You could always order the correct chimneys later on from Lehman when they are available

But all in all, it's true that an 813 needs no pressurized air when run within limits, though there still needs to be some ventilation and heat radiation flow.  IE, just think of how hot an 813 would get inside a sealed metal box with no air flow at all.

T

A friend got a  2 x 4x1 HB  old school contest amp that has cut down pickle jars for chimneys..regular glass...how cool is that?

Thanks Tom. All good suggestions to implement, as always.

Coleman lantern chimbleys have been mentioned in the past.

Just a thought,

 KLC

At 1250 VDC plate u need -160 bias in AM Plate modulation, class "C" mode. I'd use a fixed screen supply with a small 10 hy series choke for complete modulation. I like that better than no fixed bias and using a clamper tube. If you use a series resistor for screen voltage and no fixed bias, tube wastes a lot of heat and never really cuts off if keying in CW.

The bias can be derived from a VR105 plus a grid leak resistor of about 4.5 K @ 1 Watt to make -160 volts. The fixed bias is more than 2 X cutoff and resting current with no excitation should be zero. Grid current runs around 12 - 13 ma. with a drive of about 3 watts.


Seth KC2WE

Note: Been doing this since 1959 L o L

I've seen the choke in the screen supply in a couple schematics. Can anyone explain how this works, what the pros and cons are?

Ed

Looks like Seth's comment got duplicated across 2 threads...

It did, and he may have only meant the other, but it may apply here as well. The 813 data sheets say that the screen should be supplied from a separate modulated source (eg an additional winding on the mod transformer) or a resistor from the plate, for best linearity. Your design has it from a separate unmodulated source.  I see a choke in your schematics. Is the choke a good add on? What will it do in terms of the screen being modulated? And so on?

Ed

I will let Tom opine. I am mainly a kit builder here. My schematic doesn’t show that i am now planning a regulated FET 350v supply with a clamp circuit for that. I do have a winding available on the mod transformer that we could use if I get help on how to implement it.

Hi Ed,

You can use a 10H choke in series with the screen, a resistor in series with the screen or a resistor from the plate to screen of appropriate values.  It won't make much difference in modulation level or linearity which technique you use.  The amount of screen modulation compared to plate modulation is small.

Hook up a scope and SDR spectrum analyzer and run some tones thru. You can try each technique and adjust for best performance, (IMD and audio peaks) though just using a choke works out fine. The choke's impedance will vary based on audio frequency while the resistor remains constant thru the whole range.

The advantage of the choke is its DC resistance is low and the wasted power is small. When running a big tube, this can generate a lot of heat when using a resistor, especially from plate to screen.  A small tube makes it easier heat-wise.

Dean, WA1KNX wrote an article in the archives that shows how to optimize the resistor/capacitor screen network for a DX-100. But again, just throw in a 10H choke and you will be done with it for down to 30 HZ or so.

When homebrewing, I never liked a separate screen mod transformer winding to modulate the screen because you are stuck with whatever the winding ratio is and is more of a specific design for a specific tube line up, like a pair of 4-400As modulating a pair. If you use a different tube line up, then it may produce too much or too little screen modulation.

I have always used a self-modulated choke in the screen for my plate modulated 813s, 4-1000As and others rigs. I notice most of the big guns do the same thing over the years.

Tom, K1JJ

Peek A Boo

I see two!

4-1000 that is!

--Shane
KD6VXI

Show me the back side when you get it presentable.

Peek-a-boo!  That large punch that Pat N4LTA loaned me worked FB on the large meter.

Back picture is the back of the modulator. Decided to use the shielded potted filament trans on this deck. Hope that it is shielded well enough to not feed over to the driver board.

Frank has designed a new Balanced microphone preamp circuit to drive the mosfet driver. It uses the unused 6.3V windings on the mosfet driver PS to make +/- 6Vdc to run a series of op-amps to amplify and then shape the audio. Of course I did a board of it. We are awaiting sample copies to test.

John

She is starting to look nice..... I had a !Duh? moment; why is that arrow pointing at that meter?

klc

I need to find a face plate for the honking variac. That spacer between the PS and Mod will house the sequencer. I bought a nice bezel for it. Should look sharp. I had planned to have a PS control there but decided it was not needed.

Here is more on the preamp -

Frank sez - "I added a couple op amp stages of active low pass filtering. The green trace below is without the filter and the blue is with two stages. 3dB point is about 7kHz. The roll off is easily shifted with RC values. I tried to use common values for the simulation. So now we have balanced or unbalanced input with the option of gain (2dB now), a negative peak limiter and active filtering. This addresses my concerns with the wideband driver passing extreme highs. I think the layout would be cleaner with 4 discrete op amps. I only picked the OP07 because it is common and in the LT Spice library. FET op amps would be better for RF susceptibility and there are many lower noise amps out there. Sockets would allow for additional testing. The user has the option of populating the functions desired. "

Board pic... 2" by 3" - $30 for 3 of them at Oshpark.

John

Still working a lot of metal.

This punch is a beast! Makes perfect 2.75in holes.

Hooking up the variable caps to the turns counters going well. Will take a pic when farther along.

Going to make the peek holes smaller on the RF deck. Not as much space and I don't like the way the large ones worked out on the modulator. As you look down on them you see a lot of tube base. I guess they will look nice from the operating position, though.

Do these large ammeters have a shunt inside? Some of them track face current properly, some do not. I see 2 bolt heads showing under the main terminals. Will open one up to see, I guess.

John

Hi John,

That rig is going to be nice inside AND out.

I think almost every homebrewer of a big rig initially makes the mistake of not realizing that the viewing hole will not line up at certain viewing angles, like when sitting in a chair.   I have always used a single 3"  hole but after seeing yours, I think I would have done better stacking two holes for an oval shape. The 4X1s are rather large and tall, so I may get motivated to cut more to better expose the RF and modulators. I hate having to crane my neck to get the perfect alignment.

Viewing the tube plates during operation is a critical and useful requirement for both pleasure and tube longevity.  Notice the broad-cash industry did it in spades when tubes were in vogue.

T  

Looking Good! -

Got me thinking about a big rig. I found a modulation transformer.   I have 811A and 813s  I think I have some PS chokes, a plate transformer and 6 - 4uF oil caps. I have a 48 " rack

Not sure what I might use for the RF  tube. I have a used 4-1000 and Eimac air system cast socket. Not sure how good the tube is. I have a new Eimac  3CX1200A7 but I have a multi band linear planned for it. More time is what I need!



Pat
N4LTA

More pics

Wanted to show how I did the turns counters as I could find no other examples on the site. They turn real smooth and are calibrated to increase capacitance as the turns count increases.

On the top side, got the components packed up tight with minimum distances between the tank and the plate cap. Problem is that the tank bandswitch and inductor is huge. Has solid detents and should work well. Tried to wash it in the dishwasher but it still came out dingy. Time for a different approach on that. I have some copper sheet that I will cut into strips for the tank interconnect.

Waiting for a SPDT vacuum relay to deliver and have to wind a safety choke. Have 23uh on the 160M setting so thinking a 5 in long piece of 1 in pvc with around 200 wraps of 20ga magnet wire should make a pretty stout choke. Frank suggested a 10X or greater value than the 160M inductor value.

Here is the safety choke. 210uH

Yo John -

Now THAT'S a safety choke!   Built to take it and hold up during a direct HV short.  Good going.


I'm seen my share of "commercial ham" rig chokes that are fried and have opened up. They use those wimpy 2.5 mH jobs that are good at smelling and melting.  Dangerous.

You want a big wire choke in the 50 ohm load position that will take a 3KV hit to ground and  blow the 240VAC 30A breaker - and live to tell another tale.  We do not want that choke to blow during an event, like when the plate coupling cap shorts and we have HV on the antenna.

BTW, I'm building up the 3-diode high level limiter circuit for my 4-1000A plate modulated rig as a saftey net when occassionally going over 100% negative. You may want to consider it for your new 813 rig.  Use some low level limiting in the audio chain and use the 3-diode circuit to handle overshoot and mistakes that come after the the low level circuitry. Protect the mod transformer as well as keeping occassional splatter at bay.

T

A thread with the pros and cons:
http://amfone.net/Amforum/index.php?topic=38237.0

Not seeing how this translates to a Heising circuit.

It is not true Heising. Often called modified Heising.

The large inductor passes DC to the tube plate. Normally this DC passes through the modulation transformer and saturates the core, interfering with the transformer action. The core of a modulation transformer is wound on a core with an air gap which alleviates this saturation but the transformer will perform better with no DC flow. The two capacitors block DC from passing through the modulation but the AC audio current can pass.

You can use a conventionally wound audio transformer with this arrangement if the turns ratios are correct. This can be done with a large audio tube output transformer in reverse feeding a solid state amp into the speaker winding and using the high impedance winding to modulate the plate.

In a tube audio output transformer, there is no average DC current flow in the winding because the current is fed at the center tap and magnetizing current in one tube is cancelled by the magnetizing current in the other push pull tube. That happens in a push pull modulator in the primary winding also, but the secondary winding has DC only flowing in one direction.

The modulating reactor has the same thing happen, but it is wound with an air gap to help alleviate the saturation.


Pat
N4LTA

Just when we think it's safe to go back in the water, I get an email from Jeff/ NBC...

This is a link to W8JI's tests (a FB engr he is) using the 3-diode high level limiter (or variation). He feels it actually generates MORE IMD trash UNLESS it has a low pass filter after it.  This is why I like QIX's low level diode and R/C low pass filter. Without a directly coupled (DC) modulator there is a risk of overshoot from tilt by not having it high level, but I think it may be worth a try - and I will pass on the high level 3-diode circuit for now.  A high level low pass filter would be a kludgy thang too.

https://www.w8ji.com/Johnson%20audio%20mods.htm

T

I agree that he is a FB engineer.  I always go there first. I was going to add that to my transmitter, but I think I'll pass also.

BTW - You were loud and clear last night into SC. I had my 4SQRP Nouveau am rig on and  listening with it  - but a 5 watt reply was not likely to be heard.

Pat
N4LTA

Thundering along like a herd of turtles...  ::)

I don't know why it has all of those scratches on the inside. Came that way from Hammond. Annoying.

Very nice work, John....

Except for a hint of green printed circuit board, and a couple new toroid transformers, it looks retro enough to be a photograph from one of the 50's handbooks.

i like the use of gear and universal drives to make a very functional layout without detracting from optimal control locations on the front panel.  I am especially enamored of using the barrier strips for wires with lugs on the back.  This makes experimentation, customization, and measurements very easy.

Don't worry about the scratches on the underside, it is masked by the fine workmanship and planning you have invested so much time with.  It will be great to see the entire rig in a photo essay real soon!

John,

I see you made the front panel layout symmetrical, ala K1KW-style.   That took some planning.

Now keep all the RF connection short and fat - and you've arrived!

T

Let's talk about hookup wire.

For the tank, I plan on copper strips from the blocking cap through to the safety choke, then RG213 from there down to the antenna relay. The big tank inductor is already tapped about right with copper strips.

For the input side, RG213 from the connector to the cap that feeds C16. Between c16-L11-c17-c29 use strap, stranded hookup, solid? GA?

For the filaments, using #12 copper. That is the gauge of the wires coming out of the fil trans.

General connections I thought auto hookup wire (I have a bunch) that is an mix of 16 and 14 ga stranded in multiple colors. Any RF reason I should be using solid wire?

Up close on the sockets, most connections will be asap (as short as possible). For ground lugs, will put it directly to chassis.

Excellent progress!  I am also building an 813 amp and will hopefully have it up and running this week.  Thanks for sharing the details of your project!

P.S.  The last Hammond chassis I ordered also arrived with deep interior scratches.  No big deal but my OCD side would have preferred their abscence. 

Today I did a little testing. Wired up the grid 1 and grid 2 supplies. Was worried that the input voltage was too high. Loaded it down with the expected current load and the voltages came right in line. I am using PCBs that I made from circuits from AC0OB. Thanks Phil! 385V with a 80ma current limit, and a -75V regulated supply.

Most of the wiring underneath is complete. Wired a simple power supply with the 5V and 6.3V extra windings in series on the above supplies. Going to use a large 24V rotron fan running on 15V. Not loud at all. These big fans were a steal. Cost me $5 each. Plenty of CFM to keep the tubes cool. 

Now moved up top to start the wiring there. Measured all the meters. 2 are properly setup with internal shunts. 2 are not. Got online and now have calculated shunt values for them. W8JI site is really good!

https://www.w8ji.com/metering_amplifier.htm

Have plate meter wired. Will have Grid 1 and 2 current, and a voltage meter switchable to measure each grid voltage.

Next will start the copper strips to connect the plates to the tank. Ordered some AL so I can properly mount the ANT vacuum switch.

Drilled the hole and mounted the band switch knob. Need to clean up the windows. They are close. I hate hand filing. May try some door guard covering to mask the uneven cut.

Oh, well. It is home-made...

John

Is the RF choke in the backround with multiple windings insulated from ground at least 1"?    5KV under modulation can be pretty nasty.

The choke in the foreground looks OK since it is a safety choke from 50 ohms to ground.

T

Looks Great!

The single 4-1000 here uses an 8H choke in series with a 2000 Ohm resistor running on a stiff 600V supply. The screen is very nicely modulated. For fun I subbed in a 50H choke and there was little difference over the stock 8H value.

Hmmm...  I wonder how well regulated the RF final screen voltage source needs to be with a plate modulated rig?  Under normal operation the DC screen current meter stays still, but I never looked at it on a scope to see if there are any fast audio fluctuations under modulation.   My modulators have regulated screen and grid voltages, but not the RF final.


T

Mine's definitely not well-regulated, the 2K resistor in series with the 8H choke sees to that. I have bypassed the 2K resistor during testing (it is bypassed for CW mode) and there was not much difference except the 4-1000 needed less drive due to higher screen voltage, and the screen self-modulated a bit less but it was not an annoyance and the mod was still easily 100%. I did put a scope on the screen and found something like 950V peak to peak volts there, and the screen current meter was not moving much if any from the carrier level position.

The 2K res. was not the main thing, only a voltage dropper.  
The screen mod choke is where the modulation happened 8H = 25K Ohms at 500Hz.

By stiff supply, I meant just a typical choke input supply having better regulation over wide current range that a capacitor type.

I think there's tons of leeway in the self-modulating scheme but this opinion is based on only the one transmitter.

When you make the interconnects to the bandswitch, make sure you take the additional L into account.  Meaning the L from the bandswitch (minimal) and the in and output strap to and from.

This can kill efficiency.

--Shane
KD6VXI

I have not adjusted/changed the straps on the bandswitch yet. Where it is now seems close based on my plate load calculations. My plan is to wire it up fully, then feed it backwards with a nanovna at each nominal AM frequency and move the taps around for best effect. I will have the tubes installed and and some R to simulate the plate load. Is that what you are referring to with "take into account", Shane?

To add to my previous post, got the shunts worked out for the 2 grid ammeters. After calculating, I hooked up a 9V battery, an appropriate R and a digital ammeter in line with the analog one. Used an R to yield approx 3/4 deflection. Dialed it right in on the reading.

McMaster-Carr has a really good selection of rubber metal edging. And only around a $1 a foot for a 10ft quantity. I think that will clean up the window issue.

Not that far away from the final mount of all of the chassis in the rack and the final interconnect.

John

Hi John. Nice build. I think Shane is pointing to the stray L which would upset the PI match. That is something the vna test may catch as you set that network up. I think the issue may be tank L, Q degradation; adding series R. Ideal, the unloaded Q of the tank elements are very high compared to the PI match loaded Q. Curious what is your tube total plate load Z? Are you taking into account the Cout of the tube into the match design as well plate block and plate choke parasitic. Ideal, they are minor.

I am running a much smaller set of tubes. Build schematic and the vna test shown attached. I would shoot for 20 dB return loss over the band of interest. It would be interesting to see how bad the stray L attached to the band switch throws that proper tube load Z off. Not sure the vna would pick that up in measurement.

Alan  

Just noticed the flexible couplers between the gearboxes and the vacuum caps. Very nice.

I have one on that cap as I needed to transition from 1/2 diam to 1/4 diam. But it makes the rotation so smooth I am going to add one on the other side. These are only $8 at amazon. They make a real difference in releasing side tension.

Also, looks like K1JJ has found another nice upgrade to add to the RF deck. A diode limiter. Will mount it right at the plate choke base and place the indicator on the front panel.

John

Spark Gaps finalized.

Spent the day running experiments on the input pi network and the output tank.

On the output, used a MFJ-269 (a 259 with VHF/UHF) analyzer to look at the tank from the antenna back to a simulated plate load. Was able to get it to a 1.0/1.1 SWR for all bands except 160M. In talking with GFZ (Frank) he suggested that I move to a 1000pf plate coupling cap and a 1000pf padder. That did the trick and 160M came into view. I can get resonance on all bands thru 20M. I think I have too much plate capacitance to get 15M. My goal was 40 and a stretch to 20. Happy. May play a game or two to get 15M. Frank mentioned something to try.

On the input pi network, it would just not work. So I checked the roller and found that it was grounded. Apparently I hooked up the 300pf cap backwards. Swapped it and it came into resonance. Went all the way to 15M.

So, good to go for now. Awaiting some Ruskie caps that can handle the current, then will be ready for power up testing.

Now will mount and solidify the sequencer and start testing that subsystem.

My day job is in IT Healthcare, so been real busy there. Only get a few hours on the weekend to work on this. Thankfully it is a nice diversion to the real whirl.

John